An Assessment of Differential Response:
Implications for Social Work Practice in Diverse Communities
by
Amy Catherine Conley
B.A. (Boston University) 2000
M.S.W. (University of California, Berkeley) 2004
A dissertation submitted in partial satisfaction of the
requirements for the degree of
Doctor of Philosophy
in
Social Welfare
in the
Graduate Division
of the
University of California, Berkeley
Committee in charge:
Professor Jill Duerr Berrick (Chair) Professor Julian Chow
Professor Nicholas Jewell
Fall 2008
UMI Number: 3353192
Copyright 2008 by
Conley, Amy Catherine
All rights reserved.
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
®
UMI UMI Microform 3353192
Copyright 2009 by ProQuest LLC.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest LLC 789 E. Eisenhower Parkway
PO Box 1346 Ann Arbor, Ml 48106-1346
An Assessment of Differential Response:
Implications for Social Work Practice in Diverse Communities
©2008
by Amy Catherine Conley
1
Abstract
An Assessment of Differential Response:
Implications for Social Work Practice in Diverse Communities
by
Amy Catherine Conley
Doctor of Philosophy in Social Welfare
University of California, Berkeley
Professor Jill Duerr Berrick, Chair
Traditionally, the American child welfare system intervenes in cases of evident
and severe child maltreatment. Families in need of help, but who have not yet reached a
crisis level, are excluded from most government-provided family support services.
Practitioners and researchers have recently promoted the incremental development of a
complementary system. Under Differential Response (DR), families assessed as low-to-
moderate risk are referred to community-based agencies that offer voluntary, home-based
services and social service referrals. This study examined the first DR program
implemented in the state of California, at three sites in Alameda County. The research
addresses community aspects of the program's implementation, outcomes for children
and families, and staff and parent experiences with service delivery.
A mixed-methods design was used. Interviews were conducted with all
administrators (n=15), focus groups with all direct line staff (n=12), and telephone
interviews with a convenience sample of clients (n=50). Transcripts were analyzed for
emergent themes. A quasi-experimental static group design was used to examine client
outcomes. All clients who completed services formed the treatment group (n=161); a
2
comparison group was constructed with all families reported for child maltreatment in the
same timeframe who were eligible for services, but were not referred because of program
capacity (n=447). Survival analysis was used to compare rates of re-report and
substantiated re-report for the treatment and comparison groups. In Differential
Response, the community context plays a significant role. Geographic Information
Systems software was used to analyze patterns in social service availability in the three
DR target neighborhoods.
Based on interviews with staff, the program appears robust and maintains fidelity
to the model; social science theory also supports the model. However, while there was a
trend toward positive effects of the intervention, it was not statistically significant. These
findings are in line with meta-analyses of child maltreatment prevention studies, and
other studies of DR. The intervention may achieve beneficial outcomes with regards to
proximal goals including, for example, families' connections to resources, however, the
study design did not allow for examination of these effects.
TABLE OF CONTENTS
TABLE OF CONTENTS pp. i-iv
LIST OF FIGURES p. v
LIST OF TABLES p. vi
ACKNOWLEDGEMENTS p. vii-viii
CHAPTER 1: INTRODUCTION TO THE PROBLEM pp. 1-17
Scope of maltreatment in the United States
Efforts to address maltreatment
The differential response paradigm
Evaluations of differential response programs
The "Another Road to Safety" program
Organization of the Study
CHAPTER 2: MODEL DESCRIPTION, PROMISING PRINCIPLES &
LITERATURE REVIEW pp. 18-35
Background
Description of the ARS Model
Aspects of program structure and their literature base
"Voluntary" child welfare services
Home visiting
Service delivery by paraprofessionals
Current knowledge base in DR
CHAPTER 3: CONCEPTUAL FRAMEWORK pp. 36-68
ARS & the ecological model
ii
Basic needs
Attachment in parent-child relationships
Social support
Institutional resources
Conclusion: Theoretical basis of differential response
CHAPTER 4: METHODS pp. 69-81
Study site & population
Sampling
Study design
Data collection, data management, and analysis
Measures & instrumentation
Human subjects
CHAPTER 5: RESULTS (QUESTION #1) pp. 82-117
Basic needs
Critical assessment of ARS basic needs interventions
Attachment in parent-child relationships
Critical assessment of ARS parent-child relationship interventions
Social support
Perceived social support
Enacted social support
Emotional
Informational
Instrumental
iii
Critical assessment of ARS social support interventions
Connection to institutional resources
Institutional resources referral types
Institutional resources dimensions
Availability
Accessibility
Affordability
Quality & Competition
Critical assessment of ARS institutional resources interventions
CHAPTER 6: RESULTS (QUESTION #2) pp. 118-125
Why examine social service arrays?
Data collection
Need
Availability
Data management
Need
Availability
Data analysis & findings
Zip codes: Need for & availability of services
Census tracts: Need for & availability of services
Conclusion
CHAPTER 7: RESULTS (QUESTION #3) pp. 126-142
Background on statistical analysis method and hypotheses
IV
Data collection
Data management
Data analysis & findings
Descriptive statistics
Outcome: Re-report
Outcome: Investigated re-report
Outcome: Substantiated re-report
Conclusion
CHAPTER 8: DISCUSSION pp. 143-162
Limitations
Major findings
Implications for social welfare policy
Implications for social work practice
Future research
REFERENCES pp. 163-183
FIGURES pp. 184-197
TABLES pp. 198-205
LIST OF FIGURES
Figure 1: ARS Program Diagram
Figure 2: Another Road to Safety Logic Model
Figure 3: Model of stress, coping, and outcomes
Figure 4: Sampling, data collection & analysis for qualitative study
Figure 5: Sampling, data collection & analysis for outcomes study
Figure 6: Map of Alameda County, with ARS zip codes marked
Figure 7: 1 mile buffers, census tracts and zip codes
Figure 8: Zip codes—Need for & availability of services
Figure 9: Census tracts—Need for & availability of services
Figure 10: Re-report as failure
Figure 11: Investigation as failure
Figure 12: Substantiation as failure
vi
LIST OF TABLES
Table 1: State implementation of Differential Response, structural elements
Table 2: Institutional resources dimensions examined, by study
Table 3: Data on service types (total services=7,952)
Table 4: Availability of services in Alameda County, by zip code
Table 5: Comparison of demographic data, treatment and comparison groups
Table 6: Comparison of clients with re-report, investigated re-report, and substantiated
re-report, treatment and comparison groups
vii
ACKNOWLEDGEMENTS
For their support during the writing of my dissertation and throughout my
graduate studies, I am pleased to thank a number of individuals and institutions.
First, I would like to thank my dissertation committee, Dr. Jill Duerr Berrick
(chair), Dr. Julian Chow, and Dr. Nicholas Jewell, for their timely feedback and
assistance. In particular, my gratitude goes out to my advisor, Dr. Berrick, for the
mentorship she has provided on academic writing, research, grant development, and
publication.
Second, this project would not have been possible without the financial support of
several institutions. My thanks to the California Social Work Education Center, Society
of Social Work Research, Influencing State Policy, and the Society of Women
Geographers for funding my research.
Third, cooperation from the individuals and organizations involved in the Another
Road to Safety program was essential for carrying out this research. I would like to thank
the staff of Social Services Agency of Alameda County (particularly Chet Hewitt,
Michelle Love, Tom Clancy, Cynthia King Gile, and Jennifer Uldricks), First 5 Alameda
County (particularly Carla Keener, Deborrah Bremond, and Janis Burger), La Familia
Counseling Services (particularly Greta Fillingim), Family Support Services of the Bay
Area (particularly Erica Hilton), and the Prescott Joseph Center for Community
Enhancement (particularly Zakkiyyah Nazeeh).
The love and support of my parents has been the foundation of my academic
career. They have unflaggingly encouraged my ambitions and provided me with
viii
opportunities for travel and education that have shaped who I am as a person. Thank you
for believing in me.
My friends, in Berkeley and other parts of the world, have given me balance.
Meals, conversations, and fun times with Jenafir, Mary, Lisa, Bree, Jess, Nicole,
Francine, Trish, Robert, Tammi, Michelle, Anukool, and others have kept me sane.
My appreciation for Andrew Wright, who supported me during a year of
dissertation writing and job search, knows no bounds. Thanks for the tea and flexibility.
I would also like to offer my heartfelt thanks to other teachers who have mentored
me throughout the years, including Charles Lindholm, Gerald Koocher, James Midgley,
and Linda Ruth Cutts.
1
CHAPTER 1: INTRODUCTION TO THE PROBLEM
Child welfare is a high stakes field. With limited resources, administrators and
workers have no choice but to target services to those families at greatest risk of child
abuse and neglect. Yet the cost of doing nothing may be the greatest of all, if the
development and well-being of children is threatened by lack of resources and poor
parenting skills, or in the worst case scenario children die from parental injury or
negligence. Mounting research indicates that a large proportion of children initially
screened out by the child welfare hotline, or unsubstantiated after investigation,
eventually come back into contact with the child welfare system (Drake et al., 2003;
Wolock, et al., 2001; Inkelas & Halfon, 1997). Desperate families do not just go away
and endemic problems do not resolve themselves. Whether lingering at the precipice
where intervention is warranted, or tipping over to the area of full-fledged risk, the
millions of families referred yet unserved by the child welfare system represent a
population in need of help. When to intervene and what type of help to offer is a critical
issue now under discussion in the child welfare field.
Differential response is a fairly new approach to child welfare. Under the
differential response paradigm, agencies sort families by risk levels and offer services to
those deemed at moderate levels of risk, who under traditional child welfare services
would receive nothing. The differential response approach is characterized by voluntary
provision of services, greater respect for families, and increased community involvement.
This new way of doing business is catching the imagination of policy makers and child
welfare administrators throughout the country. About a dozen states have begun to
2
incorporate differential response into their child welfare systems, with California as one
of the newest states to take on this challenge.
While Differential Response offers a new perspective and changed philosophy on
how to engage and serve families, it is not an intervention per se. Agencies must decide,
based on the identified needs of their clients and available agency resources, how best to
prevent the occurrence or recurrence of maltreatment. In California, one pilot
Differential Response program called Another Road to Safety (ARS) provides an
intensive home visiting program which offers families concrete services and emotional
support. The program ultimately seeks to ensure child safety, improve child
development, and strengthen family functioning. This dissertation describes a study of
the ARS program that focused on community aspects of the program's implementation,
staff and parent experiences with service delivery, and outcomes for children and
families.
Scope of maltreatment in the United States
In 2005, 3.3 million child abuse and neglect referrals concerning 6.0 million children
were made in the United States. Slightly more than one-third of these referrals were
screened out at the hotline level, without further attention from child protective services
(CPS). Of the remaining two-thirds, more than one half (60.3%) were closed and given
the disposition "unsubstantiated" because of insufficient evidence that a child was
maltreated or at risk of future maltreatment. Only 5.6% of reported cases were given the
disposition of substantiated, meaning that evidence existed to support the finding that
children were maltreated (U.S. Department of Health and Human Services, 2007). From
the high volume of referrals, one may infer that a large number of mandated and other
reporters recognize that families need help, though their problems may not rise to the
level of statutory child maltreatment.
Researchers have in recent years begun to focus on the
substantiated/unsubstantiated distinction, questioning whether the two populations differ
significantly. The evidence is mounting that families with substantiated and
unsubstantiated allegations experience similar trajectories in terms of recurrence of child
maltreatment reporting and contact with the child welfare system. The national
prevalence of re-report among initially unsubstantiated/unfounded cases is not known.
While recurrence of maltreatment (defined as a substantiated report following an initial
substantiated report) is tracked by states and reported in the annual publication of Child
Maltreatment, current federal law does not mandate that states report statistics on initially
unsubstantiated or unfounded reports (U.S. Department of Health and Human Services,
2007).
While California, like other states, typically does not make public information on
the later child welfare trajectories of initially unsubstantiated reports, the California Child
Welfare Archive ran a series of special reports in 2005 on the request of a senior child
welfare administrator. These reports indicate that cases in which reports were initially
substantiated tend to return to the system at higher rates than other reports, but other
report types have frequent re-reports which in some instances result in substantiation. Of
all cases initially referred and substantiated between January 1, 2002 and December 31,
2002, 9.4% were re-reported and substantiated within 12 months and 13.6% within 24
months. For cases within this timeframe that were initially evaluated out without
investigation, 4.8%> were subsequently substantiated for child maltreatment within 12
4
months and 6.9% within 24 months. Cases that were investigated and given the
disposition of unfounded were re-reported and substantiated at a rate of 4.1% within 12
months and 6.7% in 24 months. Unlike many states, California also has an
"inconclusive" category; of initial reports with this disposition, 6.4% were re-reported
and substantiated within 12 months and 9.9% within 24 months. From these findings, a
pattern emerges of initially substantiated cases having the highest rates of additional
substantiated report within 24 months, followed by inconclusive cases, then evaluated
out-cases and unfounded cases at a similar rate (Needell et al., 2005).
Research on initially unsubstantiated cases comes mainly in the form of non-
population based studies which track reports and their outcomes over time. A study of
238 families in New Jersey tracked from time of first report over a period of about seven
years found that there were no significant differences in family risk factors between those
who had substantiated and unsubstantiated re-reports. In this sample, an average family
had four reports within a five year time frame, with slightly more than one-third of these
reports substantiated. Reports for a given family were random, not systematic, regarding
when a substantiated report would occur, leading the researchers to conclude that a given
report may not represent a family's risk (Wolock et al., 2001). A study of families first
reported to the Missouri Division of Family Services in 1993 and 1994 and tracked for 54
months came to a similar conclusion. Nearly half of all children (and more than half of
all cases) were re-reported to the system. Unsubstantiated victims and cases experienced
substantiation at a rate only slightly lower than substantiated cases, but made up the
largest volume of re-reported events. For three-quarters of cases in which the child was
eventually taken into care, the case was unsubstantiated during the first CPS contact
5
(Drake et al., 2003). When cases are tracked for a longer period of time, as with the
Wolock at 7 years and Drake at 4.5 years, the re-report rates appear higher than those
observed in California within a 24 month timeframe.
A longitudinal study of 12,329 investigated child maltreatment referrals in
Washington State tracked over an 18-month period found that prior CPS involvement
greatly increased the likelihood of re-referral and the rate of re-referral increased with
number of prior referrals, regardless of whether the initial referrals were substantiated or
unsubstantiated (English et al., 1999). A related study examined risk factors for multiple
referrals within close proximity to the original referral for 120,000 referrals received by
Washington State between July 1, 1994 and December 31, 1997. Time to re-referral
decreased with increasing numbers of prior reports and the time to first re-report differed
significantly from subsequent re-reports. These findings lead researchers to conclude that
that referral to CPS reflects conditions that, at least initially, increase the likelihood of
future referrals to CPS; reasons may include increased awareness among mandated
reporters in families' lives and desensitization of families' concern about CPS
intervention (Marshall & English, 1999).
Two studies examine reasons why children may cycle through the child welfare
system without receiving services. Using samples of approximately 650 Emergency
Response cases discharged in California in three time periods (1985, 1989, and 1993),
researchers found that while 50% of cases opened were given a finding of substantiated,
the majority (67.2%) of these cases were discharged after an initial investigation and risk
assessment, without further provision of services. The researchers speculate that
"recidivism may be a threshold phenomenon," with families hovering near a level of
maltreatment for which services such as family preservation or out-of-home care would
be offered by the CPS agency (Inkleas & Halfon, 1997, p. 154). A related study
examined report investigation at intake based on case reviews of three random stratified
samples (n=557) of cases from Alameda County, California. Certain types of allegations
that are more difficult to support with evidence, such as emotional maltreatment, were
more likely to be screened out. The researcher concluded that services are provided
based on legal standards of evidence rather than assessment of family's needs, and that
unsubstantiated cases may benefit from services (Lawerence-Karski, 1999).
Efforts to address maltreatment
Why do some families repeatedly come to the attention of child welfare services
without receiving an intervention? Theoretically, the system accounts for three levels of
risk. Reports with perceived lowest risk are dismissed without services. At medium risk,
Family Maintenance or Family Preservation services are provided for children who can
be maintained safely in their homes. In families whose children are at imminent risk of
harm, the highest level, out-of-home care is provided to the children and Family
• Reunification services are provided to parents.
Before considering the possible extension of in-home preventative services to
low-risk families, it is worth reviewing outcomes associated with in-home Family
Preservation services for medium risk families. Family Preservation services are often
court-mandated and attempt to address family functioning and stability so that children
may safely remain in their homes. Influenced by a variety of theoretical frameworks
including the ecological model and cognitive and behavioral theories, Family
Preservation attempts to address the multiple causes of child maltreatment.
7
Consequently, both concrete and therapeutic services are provided to families. There are
two basic program types that target different populations: Rehabilitative Family
Preservation serves families who do not present an imminent safety risk but whose
problems may require that a child be removed from the home, whereas Intensive Family
Preservation serves families where children could not be safely maintained without
provision of services or in which a family is preparing to be reunited with a child who has
been placed in out-of-home care. The latter program type, as per the name, is more staff-
intensive and time-limited, with staff carrying small caseloads so as to be constantly
available to their clients. It is also the intervention type which has been most studied,
particularly the "Homebuilder" model from Washington State (McCroskey & Meezan,
1998).
Research findings on Intensive Family Preservation have been mixed.
Randomized control studies have not found a significant difference in rates of child
placement, indicating that many families would not have entered the system in the
absence of services and that the intervention may not have been effectively targeted to
those families at imminent risk of placement. However, studies have found modest but
significant effects in other domains, including on the child's experience in the foster care
system (e.g. fewer days in placement); family functioning (e.g. supports available to
families); and child development (e.g. children's school attendance). These findings
suggest that Family Preservation can not be seen as a "panacea": better risk assessment
and case planning is needed to match services to family needs (McCroskey & Meezan,
1998). Findings from a study of Illinois' Intensive Family Preservation program support
this assertion; researchers could not identify patterns linking service components and
8
outcomes for client sub-groups, raising the possibility that the fit between client needs
and services may be more important than the breadth or array of services (Littell &
Schuerman, 2002). While still poorly understood as a construct, the importance of
tailoring services to client needs at various risk levels is clear.
There is a growing consensus among researchers that what is needed in the child
welfare system are services for the third level of risk, for those families whose allegations
of maltreatment do not meet the statutory definition or for which there is insufficient
proof, yet there is a clear need for support. The implicit reasoning behind this argument
is that, in the absence of intervention, problems in the family will escalate, degenerating
into full-blown abuse and neglect. Evidence on the similarities between substantiated
and unsubstantiated cases indicates that the finding of substantiation should not be the
gatekeeper to services (Drake et. a l , 2003). The movement towards implementing
differential response begins to fill this gap in the child welfare system. Rather than
waiting until such cases are in severe crisis and warrant coercive intervention by child
protective services, differential response offers an opportunity to engage families in
voluntary services which address their identified risk factors.
The differential response paradigm
Under the DR paradigm, a number of families who are currently screened out at
the hotline level are offered voluntary, community-based services. A DR system
involves a minimum of two pathways for families reported to the child maltreatment
hotline. Families assessed as high risk continue to receive a mandatory investigation.
Low to moderate risk families, who are generally screened out under the traditional CPS
system, are referred to community-based agencies that would assess their needs and offer
9
services. DR is comprised of three main characteristics: screening by risk level,
voluntary provision of case management and other services to lower risk families, and a
less punitive and authoritative approach than traditional child protective services. Rather
than the one-size-fits all approach that has been typical of CPS services, families under
DR receive a customized approach, with services tailored to their unique problems and
strengths (Waldfogel, 1998b).
A primary strategy for implementation of differential response involves
partnerships between child protective services and community institutions and networks
to share responsibility for serving families. Families referred to CPS often have a
multitude of needs that cannot be addressed by one agency alone. These partnerships
involve both service providers such as police, schools, and public and private agencies,
and informal helpers, such as neighborhood associations, congregations, and families
themselves. Instead of a collection of workers operating independently, families are
served by a team of community and CPS representatives who would collaborate in their
work. Depending on the level of risk for a family, either CPS or a community agency
assumes the role of lead agency for coordinating a family's services (Waldfogel, 2000).
Differential response holds great promise for families. By addressing needs early
rather than waiting until a crisis point, the psychological, cognitive, and biological
sequelae of maltreatment can potentially be averted and the family can be kept intact.
Resources can be used to provide families with parenting skills education and concrete
resources such as childcare, rather than costly payments to foster care providers and
group home facilities. The differential response paradigm makes certain inherent
assumptions about how the recurrence of child maltreatment may be prevented. First,
10
there is an assumption that risk of child maltreatment can be reduced by addressing the
presenting needs of families. As will be discussed in chapters 2 and 3, these may include
unsatisfied basic needs, such as for food or shelter, and challenges in the parent-child
relationship, related to poor bonding and attachment. Second, DR assumes that needs
may be addressed through referrals to community institutions that provide services such
as medical treatment, education, recreational opportunities, among others. Third, this
paradigm assumes that families benefit from the social support they receive from service
providers and informal helpers. Chapter 3 will examine the theoretical basis for these
intervention components.
Evaluations of differential response programs
In states with some experience providing differential response, unique approaches
have been taken in the organization and delivery of services. Case management may be
provided to lower risk families through public child welfare agencies (Missouri, Virginia,
North Carolina, Florida), through community-based agencies contracted by child welfare
(Washington, Michigan, South Carolina), or may be mixed in the state and may depend
on the county (Minnesota, Louisiana) (Schene, 2001). One worker may stay with a case
from the assessment through service delivery phase, or a case may be reassigned after
assessment. The distinct attributes which define an intervention as differential response
are: screening by risk level, voluntary provision of case management and other services
to lower risk families, and a less punitive and authoritative approach than traditional child
protective services (Waldfogel, 1998a). The varied nature of program models must be
kept in mind when interpreting research findings.
11
Evaluations of the model have been conducted in Missouri (Loman & Siegel,
2004b), Minnesota (Loman & Siegel, 2004a), Virginia (Virginia Department of Social
Services, 2003), North Carolina (Center for Child and Family Policy, 2004), and
Washington State (English et al., 2000). All of these evaluations incorporate a cohort
design for the outcomes study, and several supplement this methodology with surveys
and interviews to assess the qualitative experiences of workers, supervisors, community
members, and families. With the exception of Virginia, each of these states developed
comparison groups, either through matching a pilot and business-as-usual
county/community (Missouri, Washington, North Carolina) or through random
assignment of lower-risk families to differential response or traditional response
(Minnesota).
Each of the studies considered outcomes for children and families. Several states
also included qualitative studies which assessed organizational, behavioral, and attitude
change for the child welfare agency; community reactions; and family responses to
differential response. A cost analysis was also conducted in Minnesota (Loman & Siegel,
2004a). The main outcome measures assessed by these studies were improvement in
perceived child safety (based on observations by workers, community stakeholders, and
families), reductions in child abuse and neglect report recurrence, reductions in rates of
investigation, and reductions in out-of-home placements.
Findings from studies on differential response are equivocal. With reference to
the comparison group, families receiving differential response were statistically less
likely to be re-referred in Minnesota (Loman & Siegel, 2004a) and Missouri (Loman &
Siegel, 2004b), while no difference was observed in North Carolina (Center for Child and
12
Family Policy, 2004) and Washington (English et al., 2000). Findings for placement in
out-of-home care were mixed, with families who received differential response more
likely to have their children removed in Missouri (Loman & Siegel, 2004b), less likely in
Minnesota (Loman & Siegel, 2004a), and no difference in Washington (English et al.,
2000) (this outcome was not measured in North Carolina). Researchers for two studies
highlighted possible limitations of differential response. In Missouri, families with
chronic child abuse and neglect appeared unaffected by either differential response or
traditional services and according to researchers may have needed sustained intervention
beyond the capacity of the child welfare system (Loman & Siegel, 2004). Researchers in
Washington observed that the risk level and severity of some of the cases referred to
differential response was inordinately high, and cautioned that voluntary community
services are not designed to address severe problems (English et al., 2000).
Qualitative studies conducted in Minnesota, Virginia, and North Carolina largely
found positive opinions of differential response. Agency staff were surveyed or
interviewed in each state, while families were also interviewed in Minnesota and North
Carolina, and community partners were surveyed in Virginia. In all three states, a
majority of workers and administrators reported that the differential response system was
better than traditional child welfare services, though differential response was frequently
reported to increase workload and costs (Virginia) or present other initial challenges to
staff (North Carolina). Minnesota also reported initial cost increases, but found that
differential response was more cost-effective and resulted in a cost savings in the long-
term. Families reported high levels of satisfaction with how they were treated and the
services they received from differential response workers (Minnesota & Virginia).
13
Responses from community providers in Virginia were mainly positive or sometimes
mixed.
The U.S. Department of Health and Human Services (2005) conducted a study on
case-level data reported to the federal government (through the National Archive of Child
Abuse and Neglect Data System) for six states (Kentucky, Minnesota, Missouri, New
Jersey, Oklahoma, and Wyoming) that offer both differential response and traditional
investigation. This is the first multi-state study of differential response. Cases reported
in 2002, roughly equally divided between differential response and traditional
investigation, were compared to identify similarities and differences with regards to case
characteristics, circumstances of reports, and outcomes. Some similarities were noted in
the use of differential response. Across states, differential response was observed to be
more frequently used in cases reported by nonprofessionals or school sources (rather than
social workers, medical personnel, or legal or criminal justice sources) and cases with
allegations of neglect/medical neglect or emotional/other/unknown maltreatment.
Differential response was less likely to be used in cases in which sexual abuse was
alleged. Cases referred to differential response were more likely to be provided with in-
home services than families referred for investigation. Six months subsequent to initial
report, re-report rates appeared to be similar between those cases assigned to traditional
investigation or differential response, with the exception of Oklahoma, where rates of
subsequent reporting were lower.
Results from these studies of differential response programs have been somewhat
mixed. While rates of re-referral do not appear to be increased by participation in
differential response services, participation may increase, decrease, or have a neutral
14
effect on likelihood of child removal. This may be because the term "differential
response" encompasses a range of intervention models. A study of the ARS model
would contribute to the literature by examining outcomes for a mature model with
highly-trained paraprofessionals utilizing a home-visiting intervention, which the
literature suggests is a promising practice in the prevention of future child maltreatment
(McCurdy, 2000).
As of spring 2001, more than a dozen states had implemented or were in the
process of implementing differential response (Schene, 2001). California is a relative
newcomer to implementing the differential response model. The California Child
Welfare Redesign, a three year planning effort by 60 stakeholders to re-envision child
welfare services, recommended a shift to differential response (CWS Stakeholders
Group, 2003). The predominant version of differential response currently being
implemented in California involves three "tracks" or service responses. Moving away
from the substantiated/unsubstantiated distinction, the new approach offers services to
families based on their assessed level of risk. Track 1, called "Community Response," is
for cases that do not meet the statutory definition of child maltreatment, yet are
experiencing problems which could be addressed by services from a community-based
organization. Track 2, "Child Welfare Services and Community Response," involves a
partnership between the county child welfare agency and a community agency to provide
services for families whose reports meet the legal definition of maltreatment but in which
the risk of future child maltreatment is low to moderate and the family agrees to
voluntary participation. "Child Welfare Services Response," or Track 3, is most similar
to the traditional child welfare response, in which the county agency provides voluntary
15
or court-mandated services to families at moderate to high risk of future maltreatment
(Foundation Consortium, 2005).
The "Another Road to Safety" program
Pre-dating the California Child Welfare Services reform movement, Another
Road to Safety (ARS) has since August 6, 2002 provided a two-track differential
response. The Another Road to Safety (ARS) program uses a differential response model
to screen the risk at the county child maltreatment hotline and to offer services to families
screened out of traditional investigation who reside in certain designated zip codes in
South Hayward, East Oakland, and West Oakland. ARS clients receive up to nine
months of intensive home visiting, with services that include assistance with basic needs,
promotion of parent-child relationship development, social support, and referrals to social
services within their neighborhoods. The ultimate goal of ARS is to promote family
safety and stability to ensure positive child development.
The ARS approach to differential response resembles the predominant California
model in that there is a Track 1, for moderate to high risk cases screened out of the public
child welfare system and diverted for community services, and a Track 2, for very high
risk cases which indicate the need for court-mandated services. Since this study was
conducted, Alameda County Social Services Agency has added a third track to the ARS
program, to serve families who receive an in-home investigation but are deemed to be
low to moderate risk.
ARS has several unique attributes that make it worthy of study. First, ARS was
implemented before the CA Child Welfare Redesign, making it the first pilot differential
response program implemented in California. Second, the ARS model is unique
16
compared with differential response programs in other states and California counties with
regards to staffing (by paraprofessionals), and service delivery strategy (intensive home
visiting). Third and finally, because ARS is conducted by a different agency in each
community, it is highly tailored to the neighborhood context. As the differential response
model involves connecting families to local formal and informal resources, the ability of
agencies to form connections with other service providers and neighborhood institutions
is a key element of program design.
Data provided by the county are intriguing. Of the 1,032 families referred to the
program by February 2007, 32% were engaged in services (n=329) (C. Hwang, personal
communication). An earlier examination of the program tracked clients enrolled in the
program from May 2002-December 20004 and found that clients who received services
did not have subsequent contact with the Alameda County child welfare system in 90%
of cases (First 5 Alameda County Staff, 2004), although it is unknown whether these
families would have had subsequent child welfare contact absent the ARS program. To
date, the ARS program has not been evaluated. By now, sufficient time has passed for
the ARS program to achieve maturity and for a fairly sizable cohort of families to have
received and completed services. Evaluation to examine program outcomes in contrast to
a comparison group who did not receive the intervention is necessary to determine
whether ARS can serve as a model for implementation elsewhere in the state and country.
This study makes use of county administrative data and other data sources and
focuses on community aspects of the program's implementation, in addition to examining
outcomes for children and families, and staff and parent experiences with implementation
and service delivery. Three distinct questions are addressed:
17
Question #1: What are the experiences of ARS staff with service delivery and
clients with the services they receive, focusing on the main interventions of social support
and connection to institutional resources?
Question #2: What is the resource distribution in Alameda County and how might
institutional resources in neighborhoods influence ARS implementation?
Question #3: Is ARS successful in preventing future child welfare system
involvement?
Organization of the study
This dissertation is organized into eight chapters. Following the introduction,
chapter two provides an elaboration of the program model and critically assesses the
literature base for different aspects of the ARS intervention. In chapter three, theories
related to the two main ARS interventions of social support and connection to
neighborhood institutional resources are discussed with reference to potential for
changing child and family well-being. The study's methods are described in chapter
four. Study results are provided in chapters five, six, and seven, organized by research
question. The dissertation concludes with a discussion of the major findings and
implications for social welfare policy and practice, as well as study limitations and
directions for future research.
18
CHAPTER 2: MODEL DESCRIPTION, PROMISING PRINCIPLES & LITERATURE REVIEW
Background
The ARS program is a collaborative effort among five agencies. The program
was jointly designed by the Alameda County Social Services Agency (SSA) and
Alameda County First 5. Each California county has a First 5 commission, dedicated to
enhancing services for children under 5 and their families through use of public funds
generated from a tobacco tax. Alameda County First 5 initially funded the program, from
2002-2006, along with supplemental federal grants. Funding and oversight shifted to
Alameda County Social Services Agency in 2007. Alameda County First 5 continues to
be involved with the program through provision of database administration, training, and
case consultation. Three community-based organizations administer the program: La
Familia Counseling Service in South Hayward, Family Support Services of the Bay Area
in East Oakland, and Prescott Joseph Center for Community Enhancement in West
Oakland.
A competitive process was used to select contracted agencies for each
neighborhood, based on a review of proposals. For South Hayward, La Familia
Counseling Service was chosen as the service provider. Since 1975, the agency had been
providing culturally and linguistically appropriate mental health services to the Latino
community of Hayward. The agency's participation in the South Hayward Neighborhood
Collaborative, an association of nonprofit agencies committed to linking community
resources through capacity building and services integration, assured that ARS clients
would have access to a range of resources and services. In East Oakland, Family Support
Services of the Bay Area (FSSBA) was selected to provide services. FSSBA is based in
19
Oakland and has extensive child welfare experience, including "Family Reclaim," a
program with similarities to ARS, but targeted toward families reunifying with their
children after foster care. The Prescott Joseph Center for Community Enhancement, the
newest ARS agency, has an array of services, from health to education to arts and culture,
benefiting West Oakland.
ARS serves families reported to the Alameda County hotline for child
maltreatment who meet certain criteria (see program diagram, Figure 1): screened out of
traditional investigation and residence in the Eastmont neighborhood of East Oakland
(zip codes 94603, 94605, 94621); the Harder-Tennyson neighborhood of South Hayward
(zip codes 94541, 94544, 94545); or West Oakland (94607). These zip codes were
chosen because they have among the highest rates of child maltreatment reports in the
county. If a Social Services Agency supervisor approves the referral to ARS, the case is
forwarded to the community-based organization. When the program was initially
developed, referrals were restricted to families with a child under the age of five and/or a
pregnant mother. When Prescott Joseph began to provide services to families in West
Oakland in 2005, they began by providing services to families with children up to age 18.
For FSSBA in East Oakland and La Familia in South Hayward, eligibility was expanded
to families with children up to age 18 in January, 2007.
Description of the ARS Model
When a family is referred for services to one of the agencies, a paraprofessional home
visitor is assigned to the case. The home visitor maintains a caseload of 7-13 and sees
each family for a minimum of one hour a week. Although parents are offered services on
a voluntary basis, families who decline services are referred back to CPS for possible
20
follow-up. After conducting mental health and substance abuse screens, the home visitor
and the family jointly develop a "Family Care Plan" which outlines goals and steps to
achieve them. Goals fall under one of the following categories: Child safety; child
growth and development; parenting; school readiness; health and wellness; building
family strengths; self-sufficiency; relationships; and nutrition. These same goals are
contained in ARS's accountability matrix and are the basis for program evaluation. The
program's logic model is described in Figure 2.
With the Family Care Plan to guide the intervention strategy, home visitors have an
array of referrals they can provide for families. When clients have needs that cannot be
met through a referral, home visitors have access to a basic needs fund. Funds may, for
example, be used for food, household items, diapers, or even partial rent payments. The
concept behind the basic needs fund is to prevent the crisis of an urgent and unaddressed
need and the stress it induces.
Beyond concrete forms of help such as referrals and basic needs funds, the home visitor
develops a friendly visitor/caseworker relationship that is the intervention tool with the
family. They model healthy relationships and build trust by becoming a consistent and
supportive presence in their client's lives. Home visitors use "teachable moments" to
help parents better understand their child. This leads to improved parenting skills
because lessons are concrete, not theoretical. By helping families meet realistic short-
term goals, the home visitors hope to plant the seeds for deeper, more systemic changes
in family functioning. ARS services are offered for a relatively brief nine-month
timeframe (with three month extensions granted on a case-by-case basis), so the goal is to
21
use this period to incubate changes in parenting and life skills that will promote child and
family safety and well-being.
Aspects of program structure and their literature base
The next section will critically examine the literature base for key aspects of the ARS
program model. None of the intervention types discussed are required for the differential
response paradigm, though many have been utilized in the five states that have completed
evaluations and made their findings publicly available (Loman & Siegel, 2004a; Loman
& Siegel, 2004b; Center for Child and Family Policy, 2004; Virginia Department of
Social Services, 2004; English et al , 2000). Table 1 outlines the use of these intervention
methods in the five states which have implemented and evaluated differential response.
Without testing, it is unknown as to whether outcomes associated with the ARS program
model may be attributed to a particular intervention type, or the various interventions in
combination with the differential response pathway structure.
"Voluntary" child welfare services
ARS engages its clients "voluntarily": clients may choose to accept or refuse services,
though they are informed that in cases of refusal, child protective services will be notified
and may choose to take action. This approach differs from the usual course of action in
child welfare which involves court-mandated parent involvement in services. The child
welfare system has a "dual role structure" (Pelton, 1998, p. 127); that is, agencies hold
the responsibility of investigating maltreatment allegations and removing children who
they consider unsafe, while simultaneously promoting family preservation and offering
family support. Parents who may perceive concerns in their parenting are more likely to
22
hide from a system which bestows labels of abuse and neglect rather than voluntarily
seek out involvement (Pelton, 1998).
The ARS point of contact through the hotline report and the possibility of re-referral
to CPS should the family refuse services throws into question whether participation can
truly be considered "voluntary." While services are provided by community-based
agencies rather than child protective services workers, the specter of formal child
protective services involvement still remains. Few studies have examined the veracity of
the "voluntary" claim in child welfare services; the studies that do exist hint that some
level of coercion may still be involved. In an examination of voluntary and court-
mandated foster care services in several states, Yoshikama and Emlen (1983) found that
parents who voluntarily placed their children in foster care tended to do so for reasons of
family conflict or parental incapacitation due to illness or financial difficulties, and that
the majority reported strong influence or coercion by child welfare workers or family
members in making their decision.
What are the benefits and drawbacks of offering child welfare services on a
nominally voluntary rather than mandatory basis? Provision of voluntary services is
viewed by the field as holding promise for greater levels of client motivation (Thomas et
al., N.D.), leading to higher rates of engagement and retention in services. The field of
child welfare is just beginning to examine the concept of engagement as it relates to non
voluntary clients. From a pilot test of a multidimensional measure of client engagement
in non-voluntary child welfare services, Yatchmenoff (2005) reported findings which
indicate the presence of four underlying factors related to the latent variable of
engagement, all of which were moderately to highly correlated with each other:
23
investment in services; expectancy in the change process; receptivity to services; working
relationship between client and child welfare worker; and mistrust, an anti-engagement
dimension. Of these five dimensions, investment in services and expectancy in the
change process were so highly correlated that they were combined into a single
dimension labeled "buy-in." The dimension of buy-in had the strongest predictive
relationship to behavioral engagement, as measured by self-reported compliance with
mandated services. Based on these preliminary findings, one may cautiously infer that
community-based organizations offering voluntary services to families previously
reported to the child welfare system would do best to target those clients who are
receptive to change and do not need coercion to comply with services. Further, since
CBOs do not have the power to remove children, the interference of mistrust in the
helping process is likely minimized. However, this supposition may not hold if CBOs
have an arrangement to refer families who do not engage in services back to CPS.
Research into voluntary family support interventions has identified a host of factors at
the parent, home visitor, and community levels that influence engagement and retention
of clients. Daro and her colleagues (2005) found that initial enrollment is most
significantly predicted by intent to enroll, which in turn is influenced by the client's
readiness to change, attitude towards seeking help, and prior service experiences.
Beyond enrollment, the findings of Wagner et al. (2003) based on interviews and focus
groups from a multi-site home visiting program indicate that client engagement can occur
at different levels, suggesting that the construct of parent engagement is more complex
than merely participation or attrition.
24
Since parents can opt to leave services at any time, retention is a challenge for
voluntary family support programs. Daro et al. (2005) report that service participation in
home visiting programs is influenced by different factors at different time periods. At the
point of service engagement, the mother's perception of her infant's health risk is the
most important factor. Over time, other factors assume greater importance, including the
subjective experience of receiving services, the objective value of services received, the
characteristics of the provider and the program, and the characteristics of the community.
With regards to the community, families living in more chaotic communities were less
likely to make use of voluntary family support for extended periods of time. This finding
was replicated in a study by McGuigan et al. (2003), which found that retention for one
year in a voluntary child abuse prevention program was negatively associated with
community violence. Clearly, factors at the individual, agency, and community level
influence engagement and retention of families in family support programs. With
differential response, it will be important to gain a greater understanding of how families
perceive preventive services associated with a CPS referral, their readiness to change, and
how feelings of coercion may play a role in decisions to participate in services.
Home Visiting
Home visiting has a long history as a primary service delivery strategy with at-
risk families. The first record of home visiting as a formal social intervention in the
United States dates back to the 1880s and the Charity Organization Societies' "friendly
visitors" (Sweet & Appelbaum, 2004). In modern times, home visiting has been heralded
as an effective way to address or prevent a host of social problems; prominent supporters
include the U.S. Advisory Board on Child Abuse and Neglect and the Canadian Task
25
Force on Preventive Health Care (Bilukha et al , 2005). In light of the ARS focus on
preventing maltreatment, this section will review research findings on home visiting
programs with similar goals.
In a review of outcomes research for child maltreatment prevention programs that
used home visiting, Olds & Kitzman (1993) found that of the six studies they identified
with a randomized control trial methodology, none demonstrated a difference in child
maltreatment reports using state CPS records. However, three studies did identify
differences in rates of emergency medical services and other factors which appear to
indicate a pattern of reduced parenting dysfunction. The researchers conclude that lack
of findings using CPS records is not indicative that the programs failed to reduce child
maltreatment risk; this measure may indeed be problematic due to the greater surveillance
of participating families, which might skew reporting and inaccurately bias the rates of
reporting among participating and non-participating families.
Another meta-analysis of home visiting identified mostly positive findings for home
visitation child maltreatment prevention programs. To assess the effectiveness of home
visiting as a violence prevention strategy, Bilukha et al. (2005) conducted a systematic
review of home visiting programs that served children ages 0-2 years old and their
families and specifically measured violence outcomes in studies with a control or
comparison group. Child maltreatment subsequent to completion of services was
measured directly, through reports from child protective services, parents, or others, and
by proxy, through emergency room visits and hospitalizations for injury or ingestion,
reported injury, and out-of-home placement. Of the 21 qualifying studies (with 26
intervention arms measuring different outcomes), 20 intervention arms measured the
26
effect of home visitation on reports of child abuse and neglect by child protective services
or by home visitors; five measured the effect on rates of injury, trauma, or ingestion of
poison through medical records or mother's reports; and one measured the effect on out-
of-home placement. Members of the treatment group had lower rates of child
maltreatment than the comparison group in 19 of the intervention arms, with an overall
median effect size of -38.9%. In the remaining 7 intervention arms (of which 6
measured reports of child abuse and neglect and 1 measured out-of-home placement), the
treatment groups had a higher rate of child maltreatment than the comparison group. Of
the 21 studies included in the meta-analysis, 13 used a randomized control design.
Looking at these studies alone, the overall median effect size was -27.5% (Hahn et al.,
2005). As with the Olds and Kitzman review, Bilukha and his colleagues note that
surveillance can bias the child maltreatment report outcome. Consequently, the
researchers argue that the identified effect sizes we identified were probably
underestimates of the true impact associated with intervention.
The ability to target child maltreatment prevention through home visitation was
addressed by two meta-analyses. Sweet and Appelbaum (2004) conducted a meta
analysis of home visiting programs for families with young children. Of the 60 studies
reviewed, 18.3% were of programs with the primary goal of child maltreatment
prevention. The outcome of child abuse prevention was measured as three categories:
actual abuse (for cases reported or suspected by service providers), potential abuse (for
medical treatment that may have been associated with an incident of abuse), and parental
stress (for the potential that higher stress related to parenting may result in child
maltreatment). Child maltreatment "potential" was significantly reduced for participants
27
in programs which listed this as a primary goal, as compared to other types of home
visiting programs (Z=3.34, significant at p<.001); however, effect sizes for abuse and
parenting stress were not significant (Z=1.13 and Z=1.25, respectively). Interventions
which targeted specific populations rather than offered universal enrollment had higher
effect sizes on child cognition and potential child abuse outcomes, but lower effect sizes
for parenting behavior outcomes, a finding that the authors describe as "contradictory and
hard-to-interpret" (p. 1447). These findings highlight the challenges of assessing
outcomes of home visitation programs due to the complexity and variation of program
design, a problem that is amplified when results of many studies are combined in meta
analysis.
In a similar vein, Guterman (1999) conducted a meta-analysis to investigate the
connection between universal vs. targeted enrollment strategies and reported outcomes
for child maltreatment prevention home visitation programs. Using both measures of
maltreatment reports and parenting skills, the population-based programs (enrollment
through broadly available services systems, such as hospitals, or enrollment using
demographically-based eligibility factors, such as low socio-economic status) showed a
clear trend of greater effect-size for treatment groups as compared to screening-based
programs (enrollment based on screening for demographically-based and/or individual-
level psychosocial risk factor). One reason for this may be that programs enrolling
families based on psychosocial screens inadvertently screen-in families least likely to
change from the services offered and screen-out families more likely to benefit from the
intervention.
28
Duration of service may also contribute to client outcomes, though Olds & Kitzman
(1993) have stated their belief that quantifying number of visits and total hours of
visitation is likely less important than visit content. Findings on the impact of duration
are mixed. In the sub-sample of home visitation programs in their meta-analysis of
programs for the promotion of family wellness and child maltreatment prevention,
MacLeod & Nelson (2000) found that effect size fluctuated by number of visits, with low
effect size for programs with 1-12 visits, high effect size for programs with 13-32 visits;
low again for programs with 33-50 visits; and high again for one study on a program with
more than 50 visits. A meta-analysis of home visiting programs for families with young
children by Sweet and Appelbaum (2004) found that the differences in child
maltreatment between treatment families and controls decreased as program length
increased. In an outcomes study of a child neglect home visitation prevention program,
DePanfllis & Dubowitz (2005) found no significant difference in numbers of CPS reports
between clients randomly assigned to 3 months of intervention versus 9 months. Longer
duration of services may not result in differences among client outcomes if the
knowledge of the short duration makes staff and clients work harder to achieve goals
during the program's timeframe. Staff in programs of short duration may also make
efforts to connect families to ongoing support and services in the community that may be
similar to the types of services received by clients in programs of longer duration,
eliminating substantive differences between the interventions received (DePanfilis &
Dubowitz, 2005).
Olds and Kitzman (1993) conclude that experimental data have yet to reveal the
optimal duration or intensity of services. While the findings on home visiting as a
29
strategy to prevent child maltreatment are equivocal, home visitation has been shown to
be beneficial and to produce robust outcomes in other domains of parenting. These
include, for example, improvements in health care usage and child health in programs
targeted to low-income families.
There are differences in the observed effects of home visiting on different populations,
with often greater effects on higher than average risk groups such as low-income
unmarried teenagers. Program effects also vary by provider type; this topic will be
addressed in the next section.
Service delivery by paraprofessionals
Home visiting as a service strategy relies on the formation of a helping relationship
between the visitor and family (Wasik, 1993). Therefore, staffing of home visiting
programs is a critical component in achieving beneficial outcomes. After repeated
validation by randomized control trials, home visiting by public health nurses of at-risk
families with infants and young children (note: there are almost no studies of home
visitation of families with older children) is described as a "proven practice" by the
Promising Practices Network due to statistically significant treatment group effects on
subsequent child injuries, environmental safety, childbearing, use of public assistance,
and other health and social measures (Promising Practices Network, 2002), whereas
evidence of success by paraprofessionals is more in doubt. Yet some researchers
hypothesize that paraprofessionals may be the better candidates for home visiting to at-
risk mothers because they may better reflect the community (Sweet & Appelbaum, 2004,
Wasik, 1993), may better relate to and empathize with clients if they have also
experienced challenges as a mother (Barth, 1991; Hiatt et al., 1997), and may be able to
30
offer the types of concrete services and problem-solving approaches that clients need
(Barth, 1991). Because of these qualities, paraprofessionals may have a "reduced social
distance" with their clients (Hiatt et a l , 1997) and an easier time establishing trust
(Wasik, 1993) compared to professionals, which may aid in relationship formation and
maintenance (Hiatt et al., 1997). They may also be viewed as role models for the clients
they serve (Wasik, 1993; Hiatt et al., 1997). On the downside, paraprofessionals may
have more difficulty in achieving objectivity and setting boundaries (Wasik, 1993; Hiatt
et a l , 1997) and may not know how to intervene with families in ways that promote
mental health and self-sufficiency (Wasik, 1993).
Researchers have noted that it is difficult to quantify the effects of paraprofessional
service delivery across studies, since the term blankets a variety of individuals who differ
by educational background, training received, supervision, and duties (Hiatt et al., 1997;
Wasik, 1993). Paraprofessionals may be defined as having no post-high school education
but plenty of life experience and familiarity with the local community (Musick & Stott,
1990; Hiatt et al., 1997), or having an educational background ranging from no high
school degree to an advanced professional degree (Wasik, 1993, Wasik & Roberts, 1994).
Regardless of formal education, researchers agree that training is a critical component to
develop the necessary skills for intervention with high-risk families (Wasik, 1993; Hiatt
et al., 1997). In a national survey, with 1,492 respondents (46% response rate), Wasik &
Roberts (1994) found that of the programs employing only paraprofessional home
visitors, 43.4% reported providing in-service training, with 12.4%) of these programs
supplementing training through written materials. Seventeen percent of programs
employing only paraprofessionals reported offering no training, as compared to 47.6%o of
31
agencies employing only professionals. These findings suggest that training is perceived
as particularly important for paraprofessionals. Some form of supervision was reported
by 73% of all agencies (Wasik & Roberts, 1994). The question is: with sufficient training
and adequate supervision, can paraprofessionals provide similar services and achieve
comparable outcomes to their professional counterparts?
Two randomized control trials of paraprofessional home visiting programs for at-risk
families found minimal impact on child maltreatment outcomes. In a study by Barth
(1991), trained paraprofessionals provided six months of home visiting services to
pregnant mothers with identified risk factors for child maltreatment. At the conclusion of
services, no significant differences were found between the controls and participants in
self-reported and officially reported child maltreatment. Barth concluded that the
program's lack of success may have been due to the inabilities of paraprofessionals to
deal with the needs of highly distressed families and the short duration of services to
make long-term change in family functioning. Similarly, a randomized control trial of
Hawaii's Healthy Start, a voluntary paraprofessionally-staffed post-natal home visiting
model widely implemented throughout the United States, also found the intervention to
be ineffective in reducing rates of self-reported and officially reported child maltreatment
(Duggan et al., 2004). Duggan and her colleagues attributed the intervention's minimal
success to issues of program implementation and conceptualization.
Comparing the effectiveness of different service providers, two randomized control
trials tested differences between home visitation provided by paraprofessionals and
nurses. Korfmacher (1999) found differences between the provider types in the areas of
engagement, retention, and visit content. Compared to nurses, paraprofessionals had
32
higher passive refusal and drop-out rates. During visits, paraprofessionals spent a larger
proportion of time on environmental health and safety issues than nurses and a smaller
proportion on parenting. The researcher also noted that turnover was higher among
paraprofessional staff. A similar study conducted by Olds and his colleagues (2002)
tested the effectiveness of nurse vs. paraprofessional home visitation on maternal and
child health outcomes. Mother-child pairs served by paraprofessionals evidenced only
one statistically significant effect: mothers with low psychological resources interacted
with their children more responsively than counterparts in the group served by nurses.
For most the health and social outcomes, the effect size by paraprofessionals typically
was about half that of nurses, and effect sizes rarely achieved statistical significance.
Since the intervention was the same, the different outcomes are likely due to the type of
service provider. Nurses may have greater legitimacy and authority with the clients they
serve than paraprofessionals, particularly given health-related concerns of new parents
that may help them leverage behavioral change (Olds et al., 2002).
A meta-analysis examining the effects associated with professional vs.
paraprofessional home visitation staffing found that visitation by professionals (nurses
and mental health workers) was associated with lower rates of child maltreatment
reporting, child injury, and out-of-home placement as compared to visitation by
paraprofessionals. Program duration was also associated with effect size, with programs
of a longer planned duration more likely to produce positive results in reduction of child
maltreatment. In combination, visitor type and duration suggested strong effects;
visitation by paraprofessionals was found to be effective only in programs of two years or
longer (Bilukha et al., 2005). Another meta-analysis identified different findings, with
33
paraprofessional visitation associated with higher effect sizes than professionals in
programs addressing potential for child abuse (Sweet and Appelbaum, 2004); however,
this study utilized a weaker design by combining experimental, quasi-experimental
studies, and pre-experimental designs.
Researchers do not yet fully understand the factors associated with effective
paraprofessional home visitation. Implementation studies may help to explain some of
the mixed findings. Hiatt et al. (1997) conducted an implementation study of a home
visiting program originally designed for public health nurses and adjusted for
paraprofessionals. Program administrators recruited women without a bachelor's degree
who were mothers and older than age 18. The researchers hypothesized that due to
"shared experience" and "reduced social distance" paraprofessionals would create
relationships with their clients that differed from those of nurses. Some of the same
qualities that uniquely suited paraprofessionals for the role of home visitor (e.g. shared
experience of motherhood) created challenges for the staff in taking on this new role.
Paraprofessionals struggled with issues of gaining credibility among professional
collaborators, balancing work and home life, and adjusting to a professional culture.
Residing in the same neighborhoods as their clients brought benefits and drawbacks to
clients and paraprofessionals; while paraprofessionals could offer an insider view, they
also became hurt and defensive when professionals expressed concerns regarding the
safety of the communities. When compared to nurse implementation of the intervention
model, paraprofessionals were found to have spent twice as much time on environmental
health (e.g. safety of living conditions) and less time on personal health issues of the
mother during pregnancy. High staff turnover (50% in two years) was a problem among
34
paraprofessionals; retention was greater among those who had previous home visiting
experience. Overall, studies which compare outcomes achieved through nurse or
paraprofessional home visiting favor nurses and show negligible results for
paraprofessionals.
Current knowledge base in DR
The research literature supports certain aspects of the ARS model while casting doubt
on others. Researchers speculate that voluntary provision of services in child welfare
may enhance client motivation and better target services to those parents ready to change,
yet an often heard concern in voluntary home visitation programs is that client
engagement and retention can be challenging. Findings in the literature suggest that
families residing in chaotic and violent communities are harder to keep in programs, a
factor important to consider in programs such as ARS that are targeted to neighborhoods
with high child maltreatment rates. While services provided by the ARS program are
voluntary, they are still targeted rather than universal. Screening-based studies have
demonstrated smaller effect sizes than population-based studies (Guterman, 1999).
Home visitation is a promising intervention strategy, though the promise is not always
demonstrated (Olds & Kitzman, 1993). The research on home visiting programs that
seek to prevent maltreatment is equivocal, with meta-analyses finding no difference
related to treatment (Olds & Kitzman, 1993) as well as identifying a majority of
programs with positive effects (Bilukha et al., 2005). Different program services and
personnel configurations may explain these mixed findings. Home visiting services
staffed by paraprofessionals have achieved mixed outcomes (Sweet & Appelbaum, 2004;
35
Bilukha et al., 2005; Guterman, 2001; Barth, 1991, Duggan et al, 2004), with stronger
empirical findings for nurse home visiting.
While this literature review has not found profound evidence for all of the elements of
the ARS program structure, there are many reasons to believe that this program model
can achieve positive outcomes for the families it serves. It has a well-thought out
theoretical base (addressed further in chapter 3) that connects service inputs to expected
client outcomes. Staff members are hired with care, provided with extensive training,
and immersed in support and reflective group and individual supervision. Visitation
content is based on family empowerment and participation, with an eye to ensuring that
basic and concrete needs are met so that families can go on to address underlying
psychological and emotional problems. Family achievements are celebrated and
strengths are recognized and enhanced through services. As a promising intervention, the
next step is to empirically test the ARS model to determine its effectiveness and value for
replication at other sites.
36
CHAPTER 3: CONCEPTUAL FRAMEWORK
ARS & the ecological model
In the history of research on child maltreatment, three major models have framed
inquiry (Pecora et al., 2000). The psychodynamic model posits that the problem lay in
the individual, either the abuser or abuse victim, or in the dyad. Guided by this paradigm,
the researcher explores deviant behaviors, attitudes, and beliefs of abusive parents or the
qualities of children that are associated with abuse. The sociological model considers the
social contributors to maltreatment. Researchers influenced by this model inquire into
the connections between stress induced by poverty, among other factors, and child
maltreatment. Neither model fully accounts for the behaviors of abusive parents: some
parents with psychodynamic or sociological factors linked to maltreatment do not
exhibitive abusive behaviors (Pecora et al., 2000). A more complex model, the
ecological model, acknowledges that child maltreatment is multiply determined by
factors in the parent, child, parent-child dyad, community, and culture. No one factor is
considered predictive; the ecological model instead emphasizes that risk for maltreatment
is linked to the interaction among factors at these different levels. While the environment
influences the individual, the reciprocal is also true—the interaction is two-way. Further,
the ecological model is dynamic—factors change over the progression of a child and
family's development (Bronfenbrenner, 1979).
Ecological theory envisions individuals and their social environment as a series of
nested structures. The immediate context of the individual is the 'microsystem' and
consists of patterns of activities, roles, and interpersonal relationships in a variety of
situations. For this level, the emphasis is placed on how individuals perceive and make
37
meaning of events rather than the actual events themselves. The interrelations between
two or more settings in which the individual actively participates compose the
'mesosytem.' This may be, for example, the connections between an adult's workplace,
family, and social life, taking forms such as communication between settings and
perceptions that exist in one setting about another. The 'exosystem' is distinguished from
the mesosystem because it involves one or more settings in which the individual is not an
active participant but which affect or are affected by a setting in which the individual is
actively involved, as in the case of a school board making decisions about a child's
school. In the outermost setting, the 'macrosystem' consists of forms and content of
systems that are consistent in the lower settings, as observed in different societies and
sub-cultures, including related values and belief systems, such as values related to
parenting (Brofenbrenner, 1979). An intervention is ecological in nature if it addresses
the parent, the parent-child dyad, and the family's environment. Primary importance
must be placed on ensuring that the family has what it needs to function and provide
appropriate parenting; namely, supports such as housing, adequate health care, proper
nutrition, and employment (Brofenbrenner, 1974).
The ARS intervention, with its implicit acknowledgment of the child, parent,
family system, neighborhood, and societal contributions to the problem of child
maltreatment, fits the ecological theoretical framework of child maltreatment.
Interventions are made primarily at the micro and meso levels. The four main types of
interventions provided to families are: assistance with basic and concrete needs;
promotion of attachment in the parent-child relationship; provision of social support; and
connection to neighborhood institutional resources. The first three interventions can be
38
located within the micro level of the ecological model, whereas the last spans both the
meso-system (through influence on family systems) and micro-systems (through
interpersonal relationships with community organization staff).
Basic needs
Help with meeting concrete and basic needs is an infrequent focus of child
welfare services, even though clients have expressed desire for services that offer
material assistance (Pelton, 1982). Resolution of immediate crises can be seen as
necessary before moving on to deeper issues, such as lack of support or emotional
problems (Duggan et al., 1999). Providing parents with cash assistance and vouchers can
also offer psychological benefit to families involved in the child welfare system: having
the freedom to use cash or vouchers as the family sees fit conveys the message that the
family is valued, and that parents can be trusted to do what is in the best interests of their
children (Racino, 1998). Cash and material assistance may make a difference in those
cases where the help offered truly fits the families' needs. Indeed, help acquiring needed
equipment such as a crib, or assistance in paying a bill, may be a more effective child
maltreatment intervention than education on parenting skills or child development
(Chaffin et al., 2001). However, in cases of great financial stress, a small handout or
purchase of equipment may not tangibly improve the plight of families.
Maslow's theory on hierarchy of needs offers support for interventions that target
basic and concrete needs. Frequently envisioned as a pyramid, the Maslow (1943)
hierarchy of needs is based on the premise that "human needs arrange themselves in
hierarchies of pre-potency," (p. 3) meaning that lower needs must be at least somewhat
satisfied for higher ones to emerge. The base is constituted of 'physiological' needs for
39
food, water, and other forms of basic sustenance such as shelter and clothing. The safety
needs come next, including freedom from physical threats and security that may come
from a job and other stable routines. If physiological and safety needs are met, a desire
for love and belonging arises, with both the giving and receiving of love. Need for
esteem is the next highest level; this takes the forms of desire for achievement and
reputation or prestige. The four needs previously described make up the 'deficiency'
needs which, while unsatisfied, cause an individual feelings of anxiety. The pyramid is
capped with the need for 'self-actualization,' also described as the 'growth' need, which
consists of an individual expressing his full potential and making the most of his talents.
Motivation can be understood as dominated by the forces of deprivation and
gratification: while unsatisfied, the need for which an individual is deprived tends to
dominate until the individual is able to gratify said need (Maslow, 1970). The lowest
need tends to dominate the individual until it is sated, at which time it loses its priority to
the next need. These needs are assumed to exist in the same basic order for most
individuals. Each of these needs may be unaddressed or thwarted; for example, an
individual may have her lower needs met but be unable to fulfill the need for esteem and
therefore lack self-confidence (Maslow, 1943).
The Maslow Hierarchy of Needs is widely accepted, though supportive empirical
evidence is lacking (Wahba & Bridwell, 1976). A survey of the literature on motivation
related to employment found little support for basic tenets of the theory. The study
examined the literature related to: 1) Maslow's need classification scheme; 2) the
deprivation/domination proposition that the importance of a need is directly related to its
level of deprivation or deficiency for an individual; and 3) the gratification/activation
40
proposition that once a lower need is gratified, its importance wanes and the importance
of the next highest need increases. Factor analysis and rank-order studies that attempted
to test for the existence of the Maslow categories and their hierarchical arrangement
found no support for the existence of five underlying need categories (though in some of
the studies deficiency and growth needs clustered independently from each other) and no
support for Maslow's hierarchical arrangement of needs (though the authors caution that
rank-order is likely a poor test because the notion of conscious ranking is absent from the
Maslow theory). A group of studies testing the deprivation/domination proposition
through measures of job satisfaction and the relationship between satisfaction and job or
environmental factors found at best partial support; findings suggest that the issue of need
deprivation and domination of behavior are more complex than the theory's explanation.
Among the studies testing the gratification/activation proposition, most have
methodological flaws that render them a poor test of the theory; the best test comes from
longitudinal studies that can examine whether satisfaction of needs in one category
correlate with the importance of needs in the next level of the hierarchy over time. Still,
Wahba and Bridwell are hesitant to draw conclusions based on these studies due to their
limited time frame and examination of only two sequential needs at a time, when the
hierarchy of needs may emerge over a lifetime. There is little clear or consistent support
for Maslow's theory; however, the nature of the theory presents a challenge to empirical
testing, and extant studies had conceptual, measurement, and methodological limitations.
The authors conclude that the concept of need remains elusive, but that Maslow's theory
provides a useful departure point for generating ideas and a framework for organizing
diverse findings related to the concept of need.
41
While the preponderance of evidence from studies does not support Maslow's
theory, it continues to be frequently invoked in a number of fields (Soper, Milford &
Rosenthal, 1995). This may be because of its intuitive appeal, expressing a notion that
people can recognize from their own experiences (Simons, Irwin & Drinnien, 1987).
While the specific categories or ordering of needs may not hold in empirical testing, the
theory may yet serve as a useful heuristic for understanding the types of needs individuals
may experience, and how needs may relate to each other. Attention to what Maslow
designated as the lower, basic needs is a central part of the ARS program approach. It
may be the case that by addressing families' physiological needs of housing, food, and
other basic necessities, the ARS program allows clients to work on "higher" needs related
to safety and ultimately love. Maslow's theory also calls attention to relatedness and love
as a central human need, which is also address by the ARS program; this will be
discussed further in the next section on attachment theory.
Attachment in parent-child relationships
Attachment theory describes the formation of emotional connections among
humans. This body of theory was largely developed through the work of John Bowlby
(1969) and Mary Ainsworth (1973). Given consistent care from one or a small number of
caregivers, infants typically form attachment relationships within the first two or three
years of life. The attachment relationship provides children with protection from danger,
food, and social relationships that are a tool for making sense of the world. Four
recognizable stages of the attachment process in infancy can be identified. From birth to
2 months, infants exhibit prosocial behavior and an enjoyment for social interaction,
though they do not yet show a marked preference for their primary caregivers. This
42
changes in months 3 to 6, as infants begin to recognize particular people and show
increased preference and interaction with familiar persons. By 7 months to 3 years,
children have a selective preference for one attachment figure, which they demonstrate
through behaviors aimed at seeking out and maintaining contact with the attachment
figure, such as smiling, crying and following. From age 3 years and onward, children
develop a sense of emotional security about the attachment relationship and do not
require the same degree of physical proximity (Howe et al., 1999).
While children progress through similar stages, their attachment relationships
differ due to attributes of the parent and environment. Ainsworth advanced attachment
theory through empirical testing and identification of attachment types. Using the
"Strange Situation" procedure, Ainsworth (1978) and later researchers observed infants
in scenarios where the infant and parent are alone, the infant is alone, the infant is alone
with a stranger, and the infant is with the parent and a stranger. There are periods when
the parent enters and exits, providing opportunities to observe separation and reunion.
Based on behavioral patterns during the Strange Situation, infants can be
classified as presenting a pattern of secure attachment to the parent or a form of insecure
attachment. Securely attached infants will explore freely and interact with the stranger
when the parent is in the room, periodically checking in and using the attachment figure
as a "secure base"; when the parent departs, the infant will become upset, but will be
happy and easily comforted upon her return. Two types of insecure attachment were
identified by Ainsworth (1978): anxious-ambivalent and anxious-avoidant. Infants
characterized as anxious-ambivalent will be anxious about exploration and the stranger
when the parent is present; upon her departure, the infant will become distressed, and
43
upon her return will act ambivalently by seeking proximity but resisting comforting and
appearing resentful. Anxious-avoidant infants act disengaged from both the parent and
stranger, showing little reaction to the parent's departure or return. The work of Main
and Solomon (1986) identified an additional type of attachment: insecure-disorganized.
Behaviors associated with this characterization tend to be inconsistent and confused;
children may freeze or engage in stereotyped behaviors such as rocking when reunited
with the parent. Children may have different attachment patterns with different parents
or primary caregivers (Main & Weston, 1981).
Attachment type appears to be significantly correlated with parental sensitivity.
The primary attachment figure for securely attached infants appear adept at reading the
baby's cues and reacting in a manner that demonstrates sensitivity, acceptance,
cooperation, accessibility, and availability. Insecurely attached infants, on the other
hand, have developed defense mechanisms to cope with feelings of distress and anxiety.
The parents of anxious-ambivalent children tend to be insensitive, unreliable, and
inconsistent in their responses. To maintain contact with the attachment figure, such
children show angry behaviors, demanding attention and protection. The experience of
desiring emotional contact with an unreliable attachment figure incites feelings of
frustration and ambivalence. Children with avoidant attachment experience their parent
as rejecting, interfering, and controlling; displaying distress provokes annoyance or
agitation in their caregiver, and behaviors such as rebuffing or aggressive attempts to
control the child's behaviors. Children react by minimizing attachment behaviors such as
crying in order to stay near the parent without annoying them (Howe, Brandon, Hinings,
& Schofield, 1999). With insecure-disorganized attachment, children appear to
44
experience their parents as frightened or frightening. The insecure-disorganized child
will both approach and take flight from the parent. This may result in disassociative
behaviors (Main & Hesse, 1990). Studies of maltreated infants have identified a majority
(around 80%) as belonging to this category (Carlson, et al., 1989; Lyons-Ruth, et al.,
1991). Parents may also be the victims of trauma, which causes them to be frightened or
disassociative in the presence of their child (Main & Hesse, 1990).
Attachment patterns established in infancy have a significant influence throughout
the lifespan. Early relationships influence the development of internal working models of
social relationships. These models organize social behavior by helping an individual
make sense of the behavior of others and form expectations of what to expect, based on
past experience. While largely stable, internal working models are also dynamic and can
be modified by later relationships (Bowlby, 1988).
Attachment theory is widely accepted as the foundational theory in developmental
psychology; one that provides an integrative framework for the understanding of socio-
emotional development and personality formation through the prism of relationship
(Howe et a l , 1999). A large body of studies has been conducted to test the theory.
Prospective studies of children tested in infancy and then again in later childhood or
adolescence have largely found support for consistency of attachment style over time.
Stability of classification during the period of infancy appears highest among infants
cared for in upper socio-economic families providing quality care and lower with poorer
families providing less optimal care (Bolen, 2000). To test attachment style in adults,
researchers often use the Adult Attachment Interview. This instrument consists of thirty-
six questions covering topics related to childhood background and experiences of trauma.
45
Responses are coded for consistency of descriptions, taking into account emotional
regulation and informational content (George, Kaplan, & Main, 1985). Studies of
adolescents who were tested with the Adult Attachment Interview after being tested with
the Strange Situation protocol in infancy have found that the majority retains their
original attachment classification; with attachment style, continuity appears associated
with consistent negative or positive care and classification change with intervening life
events and changes in the relationship with the primary attachment figure (Bolen, 2000).
Though widely embraced, attachment theory is not without its critics. The
concept of working internal models and its use by attachment researchers has been
described as vague and general. The Strange Situation testing scenario has been
criticized for being culturally-specific, relying on the assumption that brief periods of
separation have the same meaning for all children though in some cultures infants are
rarely separated from their mothers. The test, which takes 20 minutes to administer, has
also been criticized for being too brief a window on the attachment relationship to assess
all its important dimensions. Coding of child behavioral patterns in four discrete
categories has also been criticized, with the assertion that attachment may instead be
better understood along at continuum. Children with abnormal relationship formation
skills, such as children with autism or those raised in institutions may not fit neatly within
one of the traditional attachment classification (Rutter, 1995).
Attachment theory suggests that sensitive and available parenting may be
promoted by certain conditions. Adequate time and a relaxed atmosphere can help
parents properly attend to their children. Parents need support themselves. In a number
of cultures, women giving birth and caring for newborns are attended by female relatives
46
who take care of chores and allow the mother to focus on her baby. A home visitor may
partially serve this role by providing "mothering" to the mother, and thereby encouraging
her to better care for her own child. Modeling of sensitive parenting behaviors by home
visitors may prove instructional (Bowlby, 1988). Parents may become more available to
their children when their stress is reduced and they experience greater feelings of
security, confidence, self-esteem, and social understanding. Children's emotional
connection with their parents may therefore be improved by interventions that offer
provision of emotional support, material support, opportunities to improve reflective
functioning through conversation, advice, and advocacy, among others (Howe, et al.,
1999). While ARS has a limited focus on parenting behaviors, there are ways in which
the program may influence attachment relationships between parents and children. ARS
creates opportunities for parents to delight in their children through special activities. By
honoring the parent-child relationship, the program may encourage greater sensitivity and
attention on the part of parents. The program also attempts to reduce stress, which may
improve parents' availability to their children.
Social support
Lack of social support or negative social support appears to be associated with
child maltreatment (Kotelchuck, 1982; Polansky et al , 1981; Polansky, Ammons, &
Gaudin, 1985). Social support is a primary function of home visiting interventions for at-
risk families because it is believed to be an effective means of leveraging behavioral
changes, skill development, and formation of connections with community resources and
service systems. As such, the relationship is a means to an end and not intended to be the
primary source of social support in a client or family's life (Eckenrode & Hamilton,
47
2000). There are two distinct approaches to the study of social support in the research
literature. The former involves activation of social support during stressful life events,
and predominates in the family stress and coping literature, whereas the latter focuses on
the role of social support in life course development/ personality and is more associated
with the field of family relations. The two approaches to studying social support are
distinct in their emphasis on the short and long-term effects of social support for
psychological functioning (Pierce et. al., 1996). The role of social support during
stressful life events will be the focus of this section.
Social support is believed to operate through perceptions of available support,
rather than actual provision of support. Perceived social support (belief that love and
support are available from significant others) appears to buffer the impact of episodic or
chronic stress and is associated with more positive health and mental health outcomes.
Perceived social support appears more correlated with personality variables than with
aspects of the social environment, and it appears to be as stable as other personality traits
when measured over time (Lakey & Lutz, 1996). Enacted social support (specific acts of
social support) has been shown to have a weak relationship with measures of
psychological symptoms, rarely demonstrating a stress-buffering effect. There is also
little evidence to document a strong connection between perceived social support and
enacted social support, as studies have consistently found only small correlations between
the two. Perceptions of others as supportive to varying degrees appear less related to
actual enacted support and more to inferences based on 'supportive' characteristics
(Lakey, et al., 1996). These findings suggest that presence of a supportive person, rather
48
than explicit supportive acts, constitute the stress-buffering effects of social support
(Thoits, 1995).
The provision of social support in relationships is theorized to consist of discrete
types of support. Based on a taxonomy developed by House (1981), the concept of social
support is most frequently operationalized as three main types: emotional, instrumental
and informational, although these categories have been criticized as being too broad and
vague for conceptual precision and empirical measurement (Barrera, 1986; Barker &
Pistrang, 2002; Jung 1987). Emotional support has the broad goal of fostering mental
health through comforting gestures intended to soothe anxiety, uncertainty, stress,
depression, and hopelessness. Interventions that characterize emotional support include
attentive listening, offering encouragement, and normalizing the situation by diverting
attention from problems. Provision of instrumental support addresses specific identified
needs. Such support involves tangible goods, such as food or shelter, or services,
including transportation and physical care (Finfgeld-Connett, 2005). Information can be
seen as the third type of social support (Hinson Langford et al., 1997), or an aspect of
both emotional and instrumental support (Finfgeld-Connett, 2005). It can take the
practical forms of facts and advice, or the more emotional forms of reassurance, empathy,
and positive affirmation. The latter is sometimes characterized as 'appraisal support' for
its role in helping to reshape self-evaluation. Informational support can be particularly
useful in times of stress, as a means of problem-solving (Hinson Langford et al., 1997).
There are shared attributes that bridge across the social support types. Social
support is context-dependent and embedded in relationships characterized by
unconditional positive regard and caring. The process of social support involves
49
advocacy, as the provider encourages the recipient to act on their own behalf and
promotes a sense of motivation through strategies such as affirmation, validation, and
encouragement (Finfgeld-Connett, 2005). There is also a presumption of reciprocity, that
help is neither unidirectional nor motivated purely out of altruism, but exchanged
(Hinson Langford et al., 1997). However, a sense of reciprocity can be cognitively
created by the provider, as with the belief that the recipient will provide help to another in
the future (Finfgeld-Connett, 2005).
Before individuals can access social support, certain antecedents must be in place.
On the part of the individual, there must be an acknowledged need for support and a
willingness to accept help from others. Others must recognize this need and have both
the willingness and ability to provide help (Finfgeld-Connett, 2005). The individual must
participate in social networks that provide the structure wherein social support may be
provided. Connections within a social network and their degree of depth and strength—
the notion of social embeddedness—are avenues of potential support. Whether the
atmosphere, or social climate, that characterizes social networks and their level of
connectedness is promotive of helpfulness and protection also plays into the availability
of support (Hinson Langford et al., 1997)
Relationships of both a personal and professional nature can provide social
support. Family and friends are often a naturally occurring, informal source of support.
More formal supports include groups and organizations within a community and
individuals such as professionals and paraprofessionals in the medical and mental health
fields (Hogan, Linden & Najarian, 2002). Support from medical and mental health
providers can be envisioned on a continuum related to level of training (Eckenrode &
50
Hamilton, 2000) from intermediate (e.g., paraprofessionals) with on-the-job training to
formal (e.g., therapists) with graduate training (Barker & Pistrang, 2002). Studies have
consistently found that people express preference for informal over formal support, and
desire professional help only when other forms of support are not available (Finfgeld-
Connett, 2005).
On the informal end of the spectrum, some providers of support may be preferred
over others. While social support generally has a positive connotation, some research
findings have suggested that there can be demands associated with 'involuntary'
relationships (e.g. relatives and co-workers) that outweigh the provision of support.
Relationships which are voluntary, as with friends or members of one's church, may
come with more manageable expectations and thereby have benefits that outweigh their
cost(Thoits, 1995).
On the intermediary and formal ends of the social support spectrum, evidence
suggests that support provided by paraprofessionals is equivalent, or perhaps even
superior, to support from professionals for mild to moderate psychological problems
(Durlak, 1979; Christensen & Jacobson, 1994; Hattie, Sharpley & Rogers, 1984). These
findings may be the result of a speculated common process of psychological helping that
produces largely equivalent outcomes across a range of treatments. Common elements of
helping relationships across provider types may include the helper's attitude (e.g.,
positive regard, empathy), efforts at persuasion, and client self-disclosure (Christensen &
Jacobson, 1994).
Similarities between client and provider may also provide a clue into the
effectiveness of paraprofessionals. Evidence suggests that match may be important, with
51
regard to who provides support and the type of support given (Thoits, 1995). When
guidance is desired and objective standards are lacking, individuals make comparative
self-evaluations and seek guidance from others who are socially and experientially
similar, when they are available. Such similarities may increase perceptions of
confidence and applicability of the guidance offered, and the likelihood that it will be
culturally acceptable. Common experiences may create a sense of empathic
understanding (Thoits, 1986). Indeed, most people seek like others in their personal and
professional environments to provide help with psychological problems rather than
formal mental health providers (Cowen, 1982). Shared experience is also the core of
self-help and mutual support efforts (Winefield, 1987). This suggests that preference for
paraprofessional or professional help may be related to congruence in class and
educational background between helper and clients. Individuals with less formal
education may prefer a paraprofessional helper of a similar background, while those with
greater amounts of formal education may prefer a professional with similar or higher
educational attainment.
Paraprofessionals may have an advantage over professionals in forming trusting
relationships with individuals and families, due to characteristics such as similar
demographic backgrounds and shared neighborhood residence. Coming from similar
backgrounds, paraprofessionals may have experienced and overcome similar stresses and,
in the process, accessed local resources, thereby having relevant experience and wisdom
which can act as the foundation for empathy. For families who feel wary or mistrustful
of professionals, paraprofessionals can provide information sharing and education from
someone perceived as a "friend." This advantage also comes with challenges, as
52
boundaries can be more difficult to maintain (Hiatt, et al., 1997). The theoretical
literature suggests that paraprofessionals may have greater success than professionals in
forming bonds and providing social support to clients of similar background, although
research on paraprofessional delivery of child maltreatment prevention services has not
always borne out this promise.
Interventions based on social support may take different approaches, depending
on theoretical orientation. One particularly salient theoretical question is whether
perception of social support is a characteristic of environments or individuals. The first
hypothesis presumes that perceived support reflects enacted social support. Interventions
with this theoretical basis frequently provide support (advice and/or reassurance) to the
persons under stress, with the expectation that such support would help the person to
cope more effectively, with fewer stress-related symptoms. This type of approach fits the
'stress-buffering' hypothesis. The second hypothesis of social support views perceived
support as a characteristic of the individual, and holds that individuals have an
interpretive bias with regards to enacted support. Interventions with this orientation often
focus on social skills development.
Interventions that provide social support from staff have not shown effectiveness
in changing naturally occurring social support networks, indicating that any protective
effects of the support are unlikely to last longer than the intervention and no enduring
stress reduction would have been provided. Success is limited to reducing immediate
symptoms among people with particular stressors (Lakey & Lutz, 1996). With regards to
home visitation models, social support absent concrete assistance in changing behaviors
has not been found effective in promoting behavioral change (Olds & Kitzman, 1993).
53
For the ARS program, these findings imply that families may be helped during the
intervention period, especially if trust has been established based on shared
characteristics and families experience a sense of 'perceived social support.' However,
without the establishment of continued social support from formal or informal sources
after termination of the intervention, sustained effects are unlikely. A crucial piece of
promoting positive parenting is helping families seek ongoing support from their
communities. The next section will consider the influence of connections to
neighborhood resources on family functioning.
Institutional resources
Local neighborhood factors including physical infrastructure, organizations that
serve residents, and relationships between the residents themselves can have a powerful
effect on family processes and outcomes (Furstenberg & Hughes, 1997). In the
ecological framework, neighborhoods can influence human development at the micro
system (through direct interactions); meso-system (through indirect influence on settings
such as the home and school); and macro-systems (through collective beliefs and norms
in a neighborhood) (Gephart, 1997). Neighborhoods may have direct or indirect effects
on children; direct through their relationships and activities with peers and other
residents, or indirect through the ways that neighborhoods influence parenting. A
number of theories have been developed on the mechanisms by which neighborhoods
influence their inhabitants (Levanthal & Brooks-Gunn, 2000; Jencks & Mayer, 1990).
Key commonalities among the frameworks are the importance of networks and resources.
In a now classic and often-cited framework, Jencks and Mayer (1990) lay out three major
54
theoretical orientations on how neighborhoods affect residents: collective socialization,
epidemic, and institutional resources models.
Collective socialization models of neighborhoods posit a mechanism whereby
neighborhoods influence children and youth through their relationships with adults and
the structures and routines that exist among residents (Jencks & Mayer, 1990; Shinn &
Toohey, 2003; Levanthal & Brooks-Gunn, 2000; Gephart, 1997). Through collective
socialization, networks may convey normative expectations for acceptable behaviors and
thereby mediate the relationship between neighborhood disadvantage and outcomes such
as adolescent prosocial behaviors (Elliott et al., 1996). Collective socialization may also
influence the level of social problems such as child maltreatment (Freisthler, 2004) and
delinquency (Kornhauser, 1978).
Epidemic models also focus on the influence of networks and transmission of
norms, but between peers rather than between adults and children (Jencks & Mayer,
1990). Behaviors are seen as a kind of contagion to which individuals have varying
degrees of susceptibility due to heredity, upbringing, or chance. The likelihood of
developing bad behaviors is related to frequency of exposure. Peer influence is
hypothesized to be strongest when other formal and informal community institutions are
lacking (Levanthal & Brooks-Gunn, 2000). Studies have found negative effects of peer
deviance on outcomes such as adolescents' school achievement (Darling & Steinberg,
1997).
Neighborhood institutional resources models (Jencks & Mayer, 1990; Levanthal
& Brooks-Gunn, 2000) hypothesize that one mechanism of neighborhood effects on
children is the presence of resources that promote stimulating learning and social
55
environments. This mechanism operates through the family, as parents must act as
advocates for their children's access to community resources (Levanthal & Brooks-Gunn,
2000). Institutional models also exert influence through relationships with adults who
come from outside the community to work in community institutions and the behavioral
regularities that come from interacting with various institutions (Gephart, 1997).
Institutional resources therefore influence children and families on the meso-system
level, by affecting the family system, as well as the micro-system level, through direct
interactions with staff of community organizations. Key community-based services for
families include: learning, social, and recreational activities; child care; schools; medical
facilities; and employment opportunities. There are four dimensions of importance
regarding institutional resources described by Levanthal & Brooks-Gunn (2000):
availability, accessibility, affordability, and quality. Competition for key resources is
another aspect of this model described by Jencks & Mayer (1990). The ARS intervention
attempts to connect families to resources in their neighborhoods. Relatively little
emphasis is placed on the connections between families and networks. This section will
parallel the ARS program by focusing on institutional resources.
Another way in which neighborhood institutions influence child outcomes is their
relative ability to help residents access resources (Small, 2006). William Julius Wilson
(1987, 1996) pointed out that neighborhood institutions, such as churches, recreation
centers, and childcare centers, may offer access to resources that may be of particular
benefit to poor or socially isolated individuals. Such institutions perform the role of
"resource broker" by creating ties to businesses, nonprofits, and government agencies in
possession of resources, and in turn giving access to these organizations and their
56
resources to individuals (Chaskin et al., 2001; Small, 2006). Resources may be defined
as symbolic or material goods that benefit individuals and include economic capital,
social capital, information, credentials, material goods, and services, among others
(Small, 2006). Involvement with "resource broker" neighborhood institutions gives an
advantage to children and families. Small argues that the "truly disadvantaged" may be
those who lack access to such institutions. Certain factors may influence why a
neighborhood institution does or does not become a "resource broker," including
"normative pressures," relating to norms of organizational functioning that may
encourage institutions or their staff to develop relationships with other providers, and
"coercive pressure" from larger authorities or flinders that mandate collaboration (Small,
2006, p. 278).
Marked differences exist between neighborhoods with regards to their
institutional resources. The primary reason for this may be the socio-economic
characteristics of neighborhoods (Jencks and Mayer, 1990; Aber et al., 1997; Pembly &
Sastry, 2003). Resources may both be less available and in greater demand in high
poverty areas. Likewise, they may be of poorer quality. When a large proportion of
residents are struggling to meet their basic needs, there may be little time left over to join
together in collective efforts to demand that politicians address local needs for services
(Fuller et al., 1997). Due to lack of financial resources, families may also not have the
option to relocate (Tienda, 1991). There is some evidence that families in resource-poor
areas are more likely to seek out and use services outside their neighborhoods (Jarrett,
1997; Coulton, 1996). Parents may see the effort of locating resources as a worthy
investment in their children's potential for success (Burton & Jarrett, 2000). Through
57
relationships with kin who live in more affluent areas, families may learn of and have
access to resources beyond their own neighborhoods. Parents may, for example, choose
to send their children to live with kin in order to benefit from better schools and
recreational opportunities (Jarrett, 1997). Usage of resources outside of one's
neighborhood presumably diminishes opportunities to develop neighborhood-based
networks of support (Coulton, 1996).
Neighborhood services may have a number of effects on children. Services
related to learning may influence a child's cognitive development. Social and
recreational activities may promote children's physical and socio-emotional well-being.
Child care can impact children's learning experiences, behavioral functioning, and
physical health. School environments may be influenced by the social and ethnic makeup
of neighborhoods, and may in turn affect children's developmental outcomes through
factors such as school quality, climate, and demographics. Access or lack thereof to
medical services may be another way that neighborhoods affect children's health and
mental health. The opportunity for employment in the community is a mediator with
particular importance for adolescents with regards to outcomes involved with the
transition to adulthood (Levanthal & Brooks-Gunn, 2000).
The level of neighborhood risk, as related to poverty, residential instability, and
childcare burden (i.e., number of available adults to care for children) appears to
indirectly affect parents' views and usage of neighborhood resources. Complementing an
aggregate analysis of Cleveland neighborhoods that connected demographic factors with
outcomes such as child maltreatment, crime, and other social problems, an ethnographic
study using in-depth interviews and neighborhood observation found that parents in high-
58
risk neighborhoods were less likely than parents in low-risk neighborhoods to use local
resources such as parks and organizations. Parents in high-risk neighborhoods frequently
expressed the belief that these places were dangerous, of poor quality, or unavailable.
Instead, they were more likely to travel outside of their neighborhoods to use such
amenities (Coulton, 1996).
The climate of local social services agencies may mirror the social problems of
neighborhoods, according to a study of two Chicago neighborhoods—one with child
abuse rates above the city average and the other with rates below, but both with similar
demographic and socioeconomic characteristics (Garbarino & Kostelny, 1993). The
attitudes of residents seem to influence local social service providers: in the
neighborhood with greater levels of child abuse reporting, a sense of negativity and
hopelessness appeared to be associated with fragmentation in the social services network
and lack of coherent community identity, whereas the more positive attitudes in the
neighborhood with lower rates of child maltreatment seemed to be related to a climate
among providers that was more supportive and interconnected. The directionality of the
relationships between child maltreatment reporting, community attitudes, and
interconnectedness among service providers was unclear. Fourteen community leaders
were interviewed in both neighborhoods, using a 12-item questionnaire developed from
prior research. In the neighborhood with higher rates of child maltreatment, the general
tone of the interviews was negative, with respondents generally unable to come up with
community strengths. Respondents neither knew much about available local services nor
gave much evidence of strong informal or formal networks of family support. By
contrast, respondents in the neighborhood with lower rates of child abuse acknowledged
59
challenges faced by the community but espoused more a hopeful view. Social service
agencies were more abundant in the latter neighborhood, better connected, and evidenced
strong formal and informal social support networks. A related study appears to bear out
the hypothesis that community agencies respond to families' needs in the neighborhood
context. Families living in neighborhoods with lower risk scores, developed on the basis
of interviews regarding family stresses and supports, were more likely to use recreational
and preventative community services as compared to families in neighborhoods with
higher risk scores, who were more likely to use treatment and rehabilitation services
(Garbarino & Sherman, 1980).
Perceptions of neighborhood resource availability may influence parental
behaviors that promote positive child development. In a study with 429 inner-city
families of African American and European American descent, African American parents
with a high sense of efficacy were more likely to engage in activities designed to develop
their children's skills and interests or provide them with positive experiences than were
European American parents with a high sense of efficacy. The researchers speculated
that differences in parental strategies had to do with differing racial perspectives on the
community environment and responsiveness of local agencies. African American parents
did not generally find communities to be supportive, and therefore had to develop
compensatory strategies, whereas European American parents could rely on more
neighborhood resources (Elder, et a l , 1995).
In a study of 368 families recruited in poor communities and schools where
children were at elevated risk for conduct disorders, researchers collected questionnaire
data on neighborhood characteristics, social networks, danger, and family context
60
variables to examine their relationship with child and parental behaviors (Pinderhughes,
Nix, Foster & Jones, 2001). Less dissatisfaction with public services (in this case, police
protection, garbage collection, the quality of schools, and public transportation) was
found to be significantly associated with consistency of parental disciplinary action and
more parental warmth in initial bivariate correlations. Greater dissatisfaction with these
public services was found to be significantly associated with more harsh parenting
interactions. In hierarchical regression analyses, dissatisfaction with public services was
found to be a unique contributor to outcomes in the case of appropriate and consistent
discipline and harsh interactions, but not with parental warmth. The authors suggest that
dissatisfaction with public services may sap the energy of parents, leaving them less able
to engage in parenting that is warm, appropriate, consistent, and non-harsh.
The majority of studies that have examined the relationship between child-level
outcomes and neighborhood-level factors have done so with cross-sectional
administrative data from sources such as the census. Consequently, these studies have
generally focused on the sociodemographic characteristics of neighborhoods rather than
their array of institutional resources (Gephart, 1997). Empirical measurement of
institutional resources are rare (Levanthal & Brooks-Gunn, 2000), and those that exist
have largely been focused on simply the presence of resources, rather than connecting
these resources back to outcomes (Sampson, Morenoff & Gannon-Rowley, 2002). A
limited number of studies that do explore or hint at the connection between child and
family outcomes and the particular types of institutional resources identified by
Levanthal and Brooks-Gunn (2000)—learning, social, and recreational opportunities;
medical facilities; schools; child care; and employment—will now be reviewed. This
61
review excludes studies that examine the connections between these resource types and
child outcomes outside the neighborhood context; for example, the effect of school
quality on children, without consideration of neighborhood factors. Table 2 describes
whether the study addresses the dimensions of availability, accessibility, affordability or
quality.
Children and families' access to and usage of learning, social, and recreational
opportunities in their neighborhood context are types of institutional resources believed
to influence a range of outcomes, though studies in this area are sparse. Participation in
after school activities may deter children, particularly adolescents, from dangerous or
anti-social behaviors and provide a chance to develop talents and strengths (Ellen &
Turner, 1997). An examination of the association between recreation centers and crime
in Columbus, Ohio census tracts revealed that the presence of recreation centers had
virtually no effect on violent crime when the level of neighborhood deprivation was very
low to moderate, but as level of deprivation worsens, the crime-reducing influence of
recreation centers increases (Peterson, Krivo & Harris, 2000). Researchers attribute this
to the social control function exerted by such institutions. From this finding, one can
speculate that, in areas of extreme deprivation, the presence of recreation centers may
reduce the risk that children and adolescents will become victims of or complicit in
violent crime.
In the Moving to Opportunity study, in which families residing in public housing
or project-based Section 8 were randomly assigned to an experimental group (with a
housing voucher that could only be used in low-poverty areas with counseling assistance
to find and adjust to new housing) or a control group (with an unrestricted Section 8
62
voucher and no counseling), children in the experimental group had lower rates of injury
(Katz, Kling, & Liebman, 2001). One reason for this may be the quality of play spaces in
neighborhoods. A quote from one mother was revealing: "The place was not safe for the
children to play. They had swings on concrete. Everything was on concrete. And that's
where most of the accidents happened" (p. 636).
While social, learning, and recreational resources are no doubt important, their
effects should also not be overestimated. A study looking at the effects of neighborhood
and family factors on early childhood (ages 1-3) developmental test scores with families
enrolled in an infant health and development program for infants with birth weight under
2500 grams (n=347) found that the availability of learning activities inside the home
outweighed the influence of learning, social, and recreational opportunities outside the
home (such as museums, parks, and libraries) (Klebanov, et al., 1998). These findings
are not surprising for children of this age; as children grow older and gain independence,
the presence of such opportunities may become more influential.
Studies on the association between health and neighborhood residence have
proliferated in recent years (Diez Roux, 2001), but access to medical facilities by
neighborhood of residence has been an overlooked area. Local medical facilities may
impact children and adults by providing routine medical care and reducing days out of
work or school, as well as treating chronic conditions such as asthma that might
otherwise remain untreated (Ellen & Turner, 1997). A study of health care use among a
cohort (n=619) of three year old children born premature with low birth-weight found
that neighborhood poverty affected health care usage controlling for family income
(Brooks-Gunn, et al., 1998). Mothers living in poor or middle-income neighborhoods
63
reported a higher number of emergency room visits than mothers in affluent
neighborhoods, possibly influencing the quality of care provided to their children.
Researchers believe this finding is most likely attributable to neighborhood
characteristics such as fewer doctors and clinics, fewer facilities with convenient hours,
and greater community acceptance of emergency room services.
While studies have looked at the effects of neighborhood stressors on school
performance (for example, Dubow et al., 1997; Garner & Radenbush, 1991), the effects
of neighborhood schools on children and families has less frequently been a focus of
study. Coleman (1968) points out that the home, neighborhood, and school environments
are the three key locations where a child may access educational resources; the need for
any one is dependent on the contributions of the other. Therefore, having access to a
good school and good teachers may be more important to children who come from
deprived home and neighborhood environments. Children are most likely to attend their
local neighborhood school, and may be more or less academically prepared and
enthusiastic about learning depending on the school's quality (Ellen & Turner, 1997).
Findings from the Moving to Opportunity study supports this hypothesis: children who
moved from public housing to both more affluent neighborhoods (experimental group)
and neighborhoods with moderate numbers of poor neighbors (Section 8 group) were
more likely to attend schools with higher achievement test pass rates than their peers who
remained in public housing, a finding which researchers attributed to a greater number
and higher quality of resources available to these higher functioning schools (Ludwig &
Ladd, 1997 in Levanthal & Brooks-Gunn, 1997).
64
Few studies have looked at characteristics of neighborhood schools on children's
educational outcomes; the impact of neighborhood socioeconomic and demographic
factors has more frequently been an object of study (Jencks & Mayer, 1990). One study
that looked at neighborhood demographics made the connection to institutional resources
as a possible mechanism. The presence of affluent neighbors was found to positively
influence three- and four-year olds' performances on IQ tests, leading the researchers to
speculate that a stronger economic base in the community is likely related to greater
availability of public and private services and community resources that contribute to
children's cognitive development (Chase-Lansdale et al., 1997). A similar finding was
identified in a study of high school graduation rates for African American children, with
high percentage of neighbors in white-collar employment associated with greater
likelihood of high school completion for African American males, controlling for family
background, early school performance, adult supervision, and substance use (Ensminger,
Lamkin & Jacobson, 1996). Indeed, the most consistent finding on neighborhood effects
on children's educational outcomes has been the positive effects of high income
neighbors on children's school readiness and achievement outcomes, particularly for
European American children (Levanthal & Brooks-Gunn, 1997). Two possible
mechanisms could be responsible for associations between the physical and social
conditions of neighborhoods and the quality of children's education in neighborhood
schools: an economic mechanism that allows for a greater number and quality of
resources, related to local tax revenues from commercial, housing, and income sources;
and the quality of staff in schools, related to the potential applicant pool available to the
65
school and the attractiveness of the school as an employer (Connell & Halpern-Felsher,
1997).
Like schools, child care is another setting highly influential to child development
that may vary by neighborhood with regards to availability and quality. Child care
facilities in poorer neighborhoods are more likely to have smaller and less experienced
staff, fewer developmentally appropriate toys and materials, and less volunteer support
from parents. Children may consequently receive less adult attention and stimulation in
more deprived settings, which may impact school readiness outcomes (Ellen & Turner,
1997). Fewer staff to care for children in lower-quality child care settings is also
associated with tolerance of aggressive behaviors by children and possibility of injury.
Conversely, high-quality child care and early intervention programs have demonstrated
positive impacts on children's cognitive and socio-emotional outcomes as well as
parenting, to some extent (Levanthal & Brooks-Gunn, 2000).
A couple of studies have identified variation in availability of child care by
neighborhood residence. A study of child care across California yielded the finding that
access to child care depends largely on geographic location, affluence, and race/ethnicity.
Some counties have a much greater supply of child care than others; for example, the
poorest areas of San Francisco had about the same quantity of child care as the most
affluent parts of Los Angeles. However, generally speaking, upper income families were
found to have the greatest access to local child care, with the middle and lower-middle
income families faring somewhat better than the poorest families with regard to access.
Across race/ethnicities, White and Black families had nearly twice the access to child
care as Latino families (Fuller et al, 1997). In New York City, a study of the locations of
66
all licensed child care centers and their association with neighborhood poverty revealed a
contradictory finding (Small & Stark, 2005). The census tracts with the highest poverty
had the highest availability of affordable licensed child care centers, particularly publicly-
funded child care such as Head Start. Neighborhood residence appears to matter with
regard to child care availability, but the relationship between location and availability
may vary regionally and may depend on state, county, and city policies (Small & Stark,
2005).
Geographic location may influence child care quality as well as availability. A
study on the quality of child care centers and family child care homes in the states of
California, Florida, and Connecticut found that quality of the child care arrangements
selected by mothers was strongly associated with geographic location. Quality of care
was assessed by factors such as providers' education levels and the intensity of structured
learning activities. Indeed, geographic location had a stronger association with quality
than most individual and family-level factors, indicating that low-income families may be
more greatly influenced by institutional forces governing the quality and availability of
child care supply than by other selection factors. Localities appear to differ in their
abilities to promote high quality child care and to make such care available to low-
income families (Fuller et al., 2004).
Employment opportunities within neighborhoods are a type of institutional
resource that is of particular importance to adolescents. Mechanisms related to
employment operate on a community-level (actual employment opportunities) as well as
an individual level (adolescents' expectations regarding opportunities for employment).
Research with low-income youth has identified a number of benefits associated with
67
employment, including access to economic resources and adult mentoring, which can
positively influence outcomes such as increased school engagement and decreased
delinquent behaviors. Attitudes adolescents develop towards employment may also
influence behaviors regarding school, substance abuse, sexual activity and fertility, and
criminal activity (Levanthal & Brooks-Gunn, 2000).
The notion that fewer jobs are available per worker in predominately Black
neighborhoods, as compared to White neighborhoods, is called 'spatial mismatch.'
According to this hypothesis, jobs have shifted out of inner cities to suburbs, creating
transportation and other barriers to employment for African Americans. A review of the
spatial mismatch literature finds strong support for the finding that spatial mismatch plays
a substantial role with regards to adolescents' access to employment. For adults, spatial
mismatch is also significant, but the research is less clear on the degree of its influence
and possible differential effects on males and females (Ihlanfeldt & Sjoquist, 1998).
Results on the social outcomes of participants in the Gautreaux Project, involving the
relocation of Chicago public housing residents to the suburbs (n=224) or within the city
(n=108) using Section 8 housing vouchers, also support the impact of residential location
on employment opportunities for youth. Youth who moved to the suburbs were more
likely to gain employment than those who remained in the city. In addition, relocated
youth were able to secure better quality jobs, as measured by earning potential (Levanthal
& Brooks-Gunn, 2000).
While empirical studies associating institutional resources types with child and
family outcomes are lacking, there is a strong theoretical basis supporting the proposition
that access to resources such as recreational facilities, child care, and medical facilities
68
can benefit children and families. For an intervention like ARS that seeks to connect
families with resources in their communities, the question is whether such resources exist
and meet the criteria of availability, accessibility, affordability, and quality.
Conclusion: Theoretical basis of differential response
Differential Response is a response path, not an intervention type per se.
Jurisdictions which choose to adopt the DR model have the flexibility to create
interventions that best address the contexts of families they serve. Nevertheless,
commonalities emerge across DR programs. Most DR programs that have been studied
make implicit use of an ecological framework and involve some combination of
interventions emphasizing basic needs, the parent-child relationship, social support and
connection to institutional resources.
In the design of this dissertation research, the theories of needs, attachment in the
parent-child relationship, social support and institutional resources will be examined with
reference to the implementation of the ARS program and the experiences of staff and
clients involved with the program. In the qualitative portion of the study, data from focus
groups with line staff, interviews with administrators, and interviews with clients is
analyzed with reference to these theories and related findings are reported. A portion of
the study looking at neighborhood resources using Geographic Information Systems
software explores the relative availability of different resource types across the three ARS
neighborhood sites. For the outcomes portion of the study, aggregate referrals to
community resources provide some backdrop to clients' trajectories of involvement with
the child welfare system.
69
CHAPTER 4: METHODS
A mixed-methods design was used for this study including qualitative and
descriptive data to answer Questions #1 & 2 and for quantitative data in response to
Question #3.
Question #1: What are the experiences of ARS staff with service delivery and
clients with the services they receive, focusing on the main interventions of attention to
basic needs, promotion of attachment in the parent-child relationship, social support, and
connection to institutional resources?
Question #2: How does the availability of institutional resources in
neighborhoods influence ARS implementation?
Question #3: Is ARS successful in preventing future child welfare system
involvement?
Study site & population
As described in chapter 2, the Another Road to Safety program is implemented in
three diverse, low-income neighborhoods in Alameda County, California. The program
is operated by a different agency in each neighborhood: La Familia Counseling Services
in South Hayward; Family Support Services of the Bay Area in East Oakland; and
Prescott Joseph Center for Community Enhancement in West Oakland. Two agencies are
involved with program administration, training, and data management: Social Services
Agency of Alameda County and First 5 of Alameda County.
The study population for question #1, on the experiences of staff and clients with
interventions provided by the ARS program, includes all current administrators, line staff,
and clients. Administrators were invited to participate either by phone or during a
70
monthly oversight meeting that involves administrators from the Social Services Agency
and First 5. Staff from the three community-based organizations were informed of the
study by their supervisors and invited to participate in two voluntary focus group
sessions. Staff invited clients to participate in the study during a routine home visit.
Current administrators, staff, and clients were selected for participation, rather than staff
involved in the past or clients who have completed services, to prevent memory bias.
Question #2, on the influence of institutional resources on program
implementation, is addressed through Geographic Information Systems (GIS) data
collection and analysis. This portion of the study involves using GIS to compare service
need (operationalized as child maltreatment reporting rates) with service availability
(operationalized as the location of services relevant to families). The population involved
in the GIS portion is all reports of child maltreatment in Alameda County, aggregated by
geographic location to the census tract or zip code. Data were available for three years
by zip code (2003-2005) and for two years by census tract (2004-2005) from the
California Child Welfare Archive. Housed at UC Berkeley's Center for Social Services
Research, the Child Welfare Archive is the repository of data on child welfare services
and clients for the state of California.
To answer question #3, on the child welfare-related outcomes of ARS clients, data
were collected on clients who were referred to and accepted services from the most
established ARS program site (La Familia in South Hayward) from May 1, 2002 to
November 15, 2007. The beginning of this timeframe corresponds to the program's
initiation. The end of this timeframe allows for the passage of nine months of ARS
services, plus a three month window of time to assess subsequent child welfare
71
involvement, for the clients with the latest enrollment dates. Analysis was restricted to
ARS-South Hayward because the researcher and ARS administrators concurred that the
program model was most established and mature, with the least staff turnover. Outcomes
for a comparison group of clients who were eligible for treatment, but not referred due to
program capacity, are contrasted to those of ARS clients. Data were drawn from the
Child Welfare Services Case Management System by Alameda County Social Services
Agency staff.
Sampling
Sampling for the qualitative study (question #1) included all current
administrators, staff, and clients involved with the ARS program (see Figure 4). The
entire population of staff was included in data collection. Interviews were conducted
with administrators at the three community-based organizations and the two oversight
agencies (n=16). Two focus groups were conducted with line staff at each of the three
agencies, with six focus groups in total, involving 12 staff members.
A non-probability accidental sample (Hoyle, Harris & Judd, 2002) was assembled
for client interviews. All English and Spanish-speaking clients enrolled from April 1,
2007 until April 1, 2008 were invited to participate by their home visitor. During a
regularly scheduled home visit, staff described the research study to clients, using a script
developed by the researcher for guidance. Staff distributed a brochure to clients, which
clearly outlined the purpose of the study and the participants' rights, as well as two copies
of a consent letter and a stamped, self-addressed envelope to return a signed copy of the
letter to the researcher. Clients elected whether or not to participate in interviews. A
total of 50 clients participated in telephone interviews.
72
Sampling for the GIS portion of the study (question #2) involved all cases
reported for child maltreatment in Alameda County over the 2003-2005 timeframe. The
entire population that fell within this parameter was included. Data were aggregated by
census tract and zip code for addresses reported.
The sample for the outcomes portion of the study (question #3) is composed of
161 clients who were referred to and retained for services with ARS-South Hayward
from May 1, 2002 to November 15, 2007 (see Figure 5) "Retained for services" means
that families at least initially agreed to participation in services, though they may not
have completed services. Only one sibling (from a sibling group) was kept in the
treatment group to preserve the statistical assumption of independence, making this a
family-level, not child-level, analysis. The entire population that met these criteria was
kept for the treatment group. The comparison group is composed of 477 cases initially
reported to the Alameda County Child Abuse hotline and evaluated-out of investigation
between May 1, 2002 and November 15, 2007. These cases met eligibility criteria for the
ARS program (child ages 0-5 and residence in South Hayward) but were not referred due
to program capacity. Like the treatment group, only one sibling was kept in the analysis.
This comparison group represents the population of cases that meet the ARS eligibility
criteria but were not referred to the program, with the exclusion of siblings.
Study design
Research Question #1 was addressed using qualitative research methods. The
experiences of administrators with program design and implementation were assessed
through in-person interviews. Interviews were guided with a standardized script of
mostly open-ended questions. Front-line staff were asked about their experiences of
73
serving families and the influence of the neighborhood context during two 1.5 hour focus
groups. A script was also used, with open-ended questions. Telephone interviews were
conducted with clients to explore their experiences with ARS services and neighborhood
organizations, as well as their experiences of raising children in their neighborhoods.
Interviews lasted approximately 30 minutes, and were guided by a script with a mix of
open and closed questions.
In Differential Response, the community context plays a significant role with
regards to the availability of institutional resources to which families may be connected.
Preliminary interviews with staff indicated that the three neighborhoods where the
program is implemented (South Hayward, East Oakland, and West Oakland) differ
significantly by demographics and social services availability. To supplement qualitative
findings on institutional resources, question #2 involves looking at neighborhood social
services arrays using Geographic Information Systems (GIS) software for data
management and analysis. Based on comments from line staff and clients about the most
commonly needed services, and the Levanthal & Brooks-Gunn (2000) framework of
institutional resources types, address data were collected on services relevant to children
and families (see Table #3). Data on neighborhood resources were geocoded to identify
the geographic coordinates for each service location, using ArcGIS 9.0 software. A layer
was built for 'service availability' using these geographic coordinates and aggregating to
the zip code and census tract levels. Service availability was categorized by quantiles as
'low,' 'medium,' and 'high,' with equal numbers of zip codes/census tracts in each
category. These categories were automatically calculated by ArcGIS 9.0 to ensure that
each contained one-third of the total services. A layer was also built for 'service need'
74
based on average rates of child maltreatment reports, by zip code and census tract, over
several years. Service need was also categorized by quantiles as 'low,' 'medium,' and
'high,' with equal numbers of zip codes/census tracts in each category. 'Availability' of
services was compared to 'need' for services by examining spatial patterns and running
correlations. Descriptive patterns of service availability and service need were analyzed
for the county, particularly with reference to ARS target zip codes.
Research to address Question #3 on client outcomes utilized a quasi-experimental
static-group comparison design (Hoyle et al., 2002). An experimental design was not
possible, because this researcher could not control assignment of clients to the treatment.
Comparison groups of referrals similar to ARS clients were instead selected. All clients
who were referred to the ARS-South Hayward program and agreed to participation (as
indicated by a signed consent form) constitute the treatment group; four comparison
groups were chosen from the California Child Welfare Services Case Management
System (CMS/CWS) and matched on month of child maltreatment report and one or both
program referral criteria: child aged 0-5 or pregnant mother in home and residence in
ARS target zip codes. While there are inherent limitations to a study in which treatment
group assignment cannot be controlled, this design maximizes the comparability of
groups by matching on referral month and program criteria. Comparison group #1,
consisting of families who met both program criteria but refused the intervention or were
unable to be contacted, controls for child age and zip code of residence, as does
comparison group #4, which consists of families eligible but not referred for ARS
services. Comparison group #2, made-up of families with children ages 0-5 who reside
in a non-ARS zip code, controls for age of the child. Composed of families who reside in
75
the ARS-South Hayward target zip codes with children ages 6-18, comparison group #3
controls for zip code of residence. By selecting four comparison groups, the statistical
power is increased because the "n" size is larger.
Data collection, data management, and analysis
Question #1 was answered through face-to-face interviews with administrators,
focus groups with line staff, and telephone interviews with clients. Interviews with
administrators and clients were typed verbatim during interview sessions. Focus groups
were either audio-taped and later transcribed, or transcribed during the focus group
session. All interview and focus group records were entered into Atlas.ti for data
management and analysis. Data analysis for staff interviews and focus groups and client
interviews involved coding for emergent themes. Records were reviewed using inductive
and deductive processes to identify major themes. Coding was conducted by the doctoral
researcher and by a graduate student researcher, to increase reliability and validity.
Regular debriefmgs were held to detect and prevent bias and negative case analysis, and
to ensure the development of an audit trail (Padgett, 1998).
Question #2 builds on the examination of institutional resources through an
analysis of spatial patterns of neighborhood resources and their relationship with child
maltreatment report rates. Data on social services locations in Alameda County were
procured from a number of sources. The data collected were the name of the agency, the
service type, and the address (some sources also included additional information not
utilized in the analysis). The most comprehensive source of data was Eden Information
& Referral, a nonprofit agency that collects data on social services and makes this
information available to the general public. Child care data were accessed through the
76
three Alameda County child care resource and referral agencies. Data on health and
mental health agencies contracted by the county were provided by Alameda County
Behavioral Health and the Health Care Services Agencies of Alameda County. The final
source of data was the internet, for data on resources such as churches and Alcoholics
Anonymous and Narcotics Anonymous meeting sites that were not otherwise available.
Zip-code level and census tract child abuse and neglect referral rates were provided by
the California Children's Services Archive, a child welfare data repository housed at U.C.
Berkeley.
Once the geographic data were acquired, they were cleaned and prepared for
analysis. ArcMap 9.0 was used for geocoding, to convert street addresses into
geographic coordinates, using the ESRI Streetmaps USA as the reference file. For
unmatched addresses, Google Maps was used to check the address in order to determine
if part of the data, such as the zip code, was incorrect and preventing a coordinate match.
An overall match rate of 99% was achieved, using these methods. Once ready, the
geocoded service locations data were joined to a zip code and a census tract file for
Alameda County, using the option "falls completely within polygon." After joining the
data, frequencies were run of the social services data for each zip code and census tract.
With the realization that administrative boundaries such as zip codes and census tracts are
artificial barriers, a 1 mile buffer was constructed around each zip and tract polygons and
all services within the buffer area were calculated. Using the frequencies with the buffer
and the area of each census tract or zip code and its buffer, a new variable of service
density was calculated.
77
Maps were constructed for the county. Maps were developed by census tract and
zip code for service availability, categorizing the service density variable as low,
medium, or high. The same was done for maps depicting service need, categorizing the
variable of average annual child maltreatment reports as low, medium, or high. These
maps were visually analyzed for trends related to service availability and service need.
Data were extracted from the GIS file format and entered into SPSS. Correlations were
run for the relationship between service availability and service need, again by zip code
and census tract.
In response to Question #3, data were drawn from administrative records in
CMS/CWS in the Alameda County Social Services Agency. The referral identifier
numbers for all families who agree to receive ARS services from ARS-South Hayward
were checked for records of contacts with the child welfare system (in the form of a re-
report) post-completion of ARS. The same was done for comparison group families who
were referred to the program in the same timeframe.
Survival analysis was used to compare the re-report rates and substantiation rates
for clients who agreed to services with ARS-South Hayward and completed services in
different time frames. For longitudinal event data, survival analysis is superior to
ordinary multiple regression in its capacity to account for censored data (for those cases
in which the event of interest did not occur in the observed timeframe) and time varying
explanatory variables (Allison, 1984). Failure events were counted beginning nine
months after referral to ARS (for the treatment group) or nine months after initial
evaluate-out report (for the comparison group). Measuring the outcome post-service
rather than post-referral minimizes the "surveillance bias" (Socolar, Runyan, & Amaya-
78
Jackson, 1995), since ARS home visitors are in their clients' homes weekly and must
report any incidents of child maltreatment as mandated reporters. The lasting effects of
service completion are also evaluated by the choice of this timeframe.
Measures and instrumentation
The dependent variables for question #1 are the main ARS interventions of
attention to basic needs, promotion of attachment in the parent-child relationship, social
support, and connection to institutional resources. These variables were operationalized
based on the theoretical framework outlined in chapter 3, which was used to develop
codes for thematic analysis.
Comments were coded for basic needs when the concept of concrete need for
materials such as food, shelter, clothing, etc. was mentioned. Codes for basic needs
included:
• Basic needs types
• Referral for basic needs
• Use of basic needs fund
Interventions focused on strengthening the parent-child relationship or creating
opportunities for parents to delight in their children were coded as promotion of
attachment in the parent-child relationship. These codes were:
• Information on child development
• Modeling of appropriate parenting behaviors
• Activities for parents and children
Social support was considered to be perceptions or acts of social support provided
by staff to clients. The following codes were developed for social support:
• Perceived social support
• Supportive characteristics of staff
• Enacted social support: Emotional
• Enacted social support: Instrumental
• Enacted social support: Informational
Connection to institutional resources was operationalized as referral to social services
assistance in following-up on service enrollment. Codes as follows were used for
connection to institutional resources:
• Institutional resource type: Learning, social, and recreational activities
• Institutional resource type: Child care
• Institutional resource type: Schools
• Institutional resource type: Medical facilities
• Institutional resource type: Employment opportunities
• Availability of institutional resources
• Accessibility of institutional resources
• Quality of institutional resources
• Affordability of institutional resources
• Competition for institutional resources
These variables were measured through focus groups with line-staff and
interviews with clients. To guide the focus groups and interviews, scripts were
developed that included questions on basic needs, parent-child relationships, social
support, and institutional resources. Both tools were developed with reference to the
80
ARS manual and in consultation with ARS staff, thereby increasing construct validity by
drawing on the same information in both cases.
The dependent variable for Question #2 is child maltreatment report (for
aggregated populations by zip code and census tract), with service availability as the
independent variable. The independent variable is operationalized as those services most
frequently used by ARS families, according to focus groups with staff, and those
resources hypothesized to improve child and family outcomes, according to institutional
resources theory (Levanthal & Brooks-Gunn, 2000).
The dependent variable for Question #3 is involvement with the child welfare
system for families who accepted ARS services, compared to similar families. The
variable is operationalized as a re-report after 9 months of ARS referral (or initial
evaluated out report for the comparison group), investigated re-report, and substantiated
re-report. The independent variable is acceptance of ARS services, by both agency and
family. Other factors, such as child's ethnicity, gender, and number of prior reports, were
examined as potential confounders of treatment effects.
Human subjects
Approval for this study was secured from Berkeley's Committee for the
Protection of Human Subjects. Permission for the qualitative portion of the study was
provided on July 21, 2006 (CPHS Protocol #2006-5-21). Renewal for the qualitative
study and permission for the outcomes study was provided on August 8, 2007 for data
collection and analysis through August of 2008 (with the same protocol number).
Measures were taken to protect human subjects through the data collection, data
management, and analysis phases of the study. For the qualitative study, client
81
participation in interviews was voluntary. Clients were invited to participate in
interviews by their home visitor. Home visitors also distributed literature on the research
project, in the form of a brochure describing the study and the participants' rights and a
letter of consent. Clients were told verbally and in writing that refusal to participate
would not affect service delivery and that their comments would be kept as confidential
as legally possible. Prior to interviewing ARS staff, administrators, and collaborators, the
researcher verbally informed them as to the voluntary and confidential nature of their
comments. Quotes from ARS clients or staff that appear in this dissertation or related
publications are not attributed. Consent forms from clients and staff were kept in a
locked file cabinet at the Center for Child and Youth Policy. No human subjects issues
were raised for the map development, as all sensitive child welfare data were in aggregate
and non-identified forms. For the outcomes study, client identifier numbers and records
of child welfare histories were kept on an external hard-drive on a non-networked
computer and stored in a locked filing cabinet when not in use, as were hard copies of
client information.
82
CHAPTER 5: RESULTS (QUESTION #1)
ARS programmatic interventions focus on four primary domains: basic needs,
attachment in the parent-child relationship, social support, and connection to institutional
resources. In practice, these categories are not always distinct; specific interventions may
blur the line. For example, assistance connecting with a food bank may fall within the
category of basic and concrete needs as well as connection with institutional resources.
Findings from interviews with managers, focus groups with line staff, and interviews
with clients will now be discussed and organized around these four themes.
Basic needs
Attention to basic needs takes precedence over other types of interventions, according
to staff. A number of ARS workers suggested that families first request help to address
basic needs, and that these must be addressed before moving on to deeper issues. A staff
member described this as "a progression from basic to deeper needs." One clinical
supervisor referred to the Maslow hierarchy of needs, and explained that many families
are at a lower level of the pyramid, with unmet needs for food and shelter. Needs such as
these can not be ignored or bypassed; as one clinical supervisor suggests:
To build the relationship and try to meet people's needs, you have to seem real to them and if you're not at every place that they are, struggling with the issues with them, if you can't help them with those issues, then it's hard for them to feel supported to do work and make change.
Identifying families' unmet basic needs and offering immediate assistance is
crucial to the engagement process described by all three agencies. Staff ask families
what they need, and explain what they can offer. In the case of ARS-East Oakland, this
takes the form of an offer to help supply the family with basics like food, as well as
services and referrals. At the very first visit, the ARS-East Oakland clinical supervisor
83
mentioned that they will bring diapers for families with very young children and school
supplies for older children. She suggested that this thoughtfulness around family needs
promotes connection with the home visitor. West Oakland staff also offer to help with
housing, food, and clothes as part of their initial pitch to newly referred clients. ARS-
South Hayward staff stated that they offer to help families with whatever it is that they
need during the initial visit; the clinical supervisor emphasized that they leave families
with the "promise of some tangible help." Helping families to address basic needs in a
timely fashion may help facilitate engagement in services and cement the relationship
between home visitor and client.
Many of the most typical types of family needs described by staff and
administrators may be considered basic needs. When asked to list the most important
needs faced by families, staff at all three agencies mentioned housing and employment.
Families often live in substandard or overcrowded housing, staff report, and parents are
frequently unemployed. Shelter falls under the physiological needs on the Maslow
hierarchy, while employment falls under the safety needs by providing financial security
and regular routines. Other physiological needs of families mentioned by staff include
those for food, clothing, and basic health care. Additional safety needs cited by staff
were protection from neighborhood and domestic violence. Needs at higher levels of the
hierarchy (for love and belonging, esteem, and self-actualization) were mentioned less
frequently, though staff in West Oakland emphasized that families lack in consistent
support from the community.
ARS workers have a basic needs fund available to address one-time, acute needs;
however, two of the three program sites have chosen to reduce this pot of money in order
84
to divert funds to other aspects of the program. Staff use the basic needs fund to pay a
late electricity bill, provide a box of food, or purchase a bag of diapers. In summer
months, staff often use the fund to pay for recreational activities or summer camp. In
some cases, workers will even help clients with a rental deposit or a car repair bill. Staff
at one agency stated that they will occasionally purchase monthly bus passes for their
clients. Addressing acute needs with an infusion of cash may be a critical intervention.
Staff at one agency cited the basic needs fund as one of the most important types of
assistance provided by the program.
While staff sees assistance with basic needs as an important service to families,
they also ensure that clients will be able to address their own needs in the future. For
example, if an ARS worker pays for a late bill out of the basic needs fund, they work with
the family to find out what is keeping them from paying their own bills. Staff points out
that building self-sufficiency is a central goal of the program. One staff member
explained:
Because the point of us being there is to help them kind of access resources, but we have to leave. So when we leave we need them to be able to rely on something or be able to go back there and access [resources].
Whenever possible, staff make referrals to community organizations rather than pay for
goods and services directly. For example, rather than purchasing furniture staff may first
try to solicit a donation from an organization that provides free furniture. After helping a
family pay a high electricity bill, staff state that they will then assist with enrollment in
programs providing discounts to low-income families. Above all, staff help families
learn how to use resources in their communities so that they may return to these
85
organizations in times of future need. The process of referring families to services is
further elaborated in the institutional resources section.
Clients frequently reported receiving assistance with meeting basic needs. These
included assistance with food (e.g., food boxes or bags of groceries, food stamps, and
special food for holidays such as Thanksgiving); clothing (e.g., for school); necessities
for children (e.g., diapers, toys, and books); basic medical treatment (e.g., assistance in
acquiring health insurance through MediCal and referrals to clinics); housing (e.g.,
providing housing listings and helping with applications for Section 8); employment
(e.g., assistance with development resumes and print-outs of job listings); and income
support (e.g., applications for CalWorks). A few clients mentioned that their worker will
ask if the family needs basic things like food and diapers during home visits, and will
respond quickly when such requests are made.
Several clients expressed their appreciation for assistance with basic needs,
particularly with food and diapers. This kind of help, according to one mother, "took the
pressure off and allowed her to make necessary changes in her life. One client reported
that the offer to assist with basic needs encouraged her to accept the program:
I have had many problems. I sometimes didn't even have food or clothes for my children. I was really depressed. I felt like I wasn't worth anything. But when [ARS worker] came to my house, she told me that they can help.
Some families received assistance with moving into a new home, which frequently had a
major effect on their lives. One client noted that with assistance from ARS, she was able
to move from a cramped and overcrowded apartment in a dangerous area to a suburban
home with more space and better access to good schools and parks. Another client
mentioned receiving financial assistance from the basic needs fund to pay rent when the
86
father in the family was not working. When asked how ARS could be improved, one
client emphasized a focus on basic needs: "Just to make sure that they [the families] have
resources like bus passes, diapers, food... Stuff that we really need."
Critical assessment of ARS basic needs interventions
The ARS intervention reflects the tenets of Maslow's hierarchy of needs theory
by its attention to addressing unmet basic needs, particularly physiological and safety
needs, during the initial engagement process and throughout the nine month invention.
Fulfillment of basic needs is acknowledged by staff to be a necessary pre-requisite to
working on parenting abilities, related to the higher need for love and belonging. Clients
expressed appreciation of this type of intervention and many acknowledged that
assistance from ARS in meeting family crises such as lack of food, substandard housing,
and unemployment helped them better care for their children.
This study design did not test families for their presenting needs vis-a-vis the
Maslow Hierarchy of Needs. As described in chapter three, empirical testing of needs
classification has typically done a poor job of capturing the categories included in the
Maslow schema, either because the theory or testing methods or both are lacking. The
most that can be said, based on the client interviews, is that some clients reported feeling
calmer (i.e., less anxious) and more confident in their parenting once their basic needs
were met and family crises at least partially resolved. When asked what had changed in
herself or her family after the ARS intervention, one client stated: "I'm more calm with
my children, [better] able to deal with them." This mother had previously explained that
her home visitor had been responsive in helping the family meet all the needs they had
discussed, including for food and new housing. Another client, when asked about
87
changes in her life due to the ARS intervention, stated that there had indeed been
changes: "Yes, in my life... because before I was always stressed because I couldn't buy
this or that for my children...But now with the support, I feel more calm. Like when I
don't have any diapers, Rosalinda does what she can to help me get them. I feel more
calm and more reassured." Similar comments were made by other clients. These
statements appear consonant with Maslow's presumption that a person will feel anxious
about an unmet need until it is filled. Statements by clients also suggest that ARS helped
them to meet lower physiological and safety needs and therefore allowed them to begin
attending more to parenting and the related need for love and belonging.
The ARS approach generally emphasizes the two lowest levels of the Maslow's
hierarchy, with some degree of attention to the third level of need for love and belonging,
primarily through the work on the parent-child attachment relationship. Maslow's theory
acknowledges that lower needs can regain primacy if they go unfulfilled after a period of
being satiated. For this reason, the ARS focus on offering families information and skills
to independently access services may allow families to maintain the gains they make
during the program. While assistance from the basic needs fund is limited to the
intervention period, the ability to access resources can help families sustainably meet
their needs long after they have completed the program. This could conceivably allow
families to continue to focus on their parenting and other higher needs, if lower basic
needs continue to be met. Efforts by ARS staff to help families with parenting issues will
be addressed next.
88
Attachment in parent-child relationships
Interventions related to the parent-child relationship were acknowledged by staff at
all three agencies to be a sensitive issue. The degree to which staff focus on the parent-
child relationship varies by agency. Related interventions include: offering general
information on child development, giving specific advice related to families' issues, and
providing opportunities for parents and children to spend fun time together.
ARS-East Oakland staff expressed feeling somewhat hesitant about making
interventions related to parenting. This is in part due to concern that staff present the
program as primarily providing connections to community resources and that parents
might be offended by interactions emphasizing the parent-child relationship. Another
reason is their own limited training as paraprofessionals. One staff member argued that
the ARS staff were not qualified to intervene in attachment and bonding, as a specialist in
child development would be. In her opinion, while the parent-child relationship is a
major focus for First 5 Alameda County, ARS-East Oakland staff see the program's
purpose as dealing with more basic family problems and are doubtful that their clients
would be interested in working on the child's development and the parent-child
relationship if the child is developing normally.
Staff at ARS-South Hayward appear to be more willing to intervene in the parent-
child relationship when they feel that parents are open to such advice. They first spend
time observing interactions between parents and children, attuned to signs that might
indicate a problem; from First 5 trainings they have learned the signs of healthy
relationships as well as red flags. Family science classes organized by ARS-South
Hayward and the Lawrence Hall of Science (further described later in this section)
89
provide an opportunity to see whether parents are actively involved with their children or
more removed. If ARS-South Hayward staff sense that parents are receptive and
interested in help, they will provide parenting advice themselves or arrange for an in-
home consultation with a First 5 child development expert. Staff acknowledged that
some parents have issues with trust, making intervention more difficult. The process of
offering parenting advice can be facilitated by completing child development assessment,
such as the Ages and Stages Questionnaire. Having an assessment to refer to, rather than
relying on observation alone, "makes the points more valid," suggested one staff
member.
Unlike the ARS programs in East Oakland and South Hayward, which were
initially focused on serving target children ages birth to 5 years old, the West Oakland
program was originally conceived and implemented to serve target children up to age 18.
This has implications for the way that ARS-West Oakland approaches the issue of the
parent-child relationship. Unless parents initiate a discussion, ARS-West Oakland
workers will wait until trust is established and basic needs addressed before delving into
parent-child relationship issues. Once the timing is right, staff described taking a
customized approach depending on the child's age. Work with families who have
children under the age of five typically involves developmental assessments and
education on how to promote their child's healthy development. For families with
school-age children, the focus is often on the schools and ensuring that parents
understand their rights. With teenagers, staff help to negotiate the parent-child
relationship and diffuse tensions. Staff comment that some of their work promotes
bonding, by providing families with tickets to activities and events so that they can enjoy
90
time together, but more frequently their work is directed toward helping families manage
crises and trials that face them.
ARS-South Hayward clients seemed to comment on receiving interventions
related to parenting most frequently, but a number of clients served by the other agencies
also described parenting assistance provided by their workers. Parenting-related
assistance described by clients primarily took the form of provision of child development
information, discussions of parenting challenges, and suggestions for different parenting
techniques. Clients reported receiving information on topics including children's
developmental milestones, nutrition, baby-proofing, and activities to do with small
children. A couple of clients referred to the developmental assessments conducted by
their ARS worker, and referrals to developmental specialists at local public and private
agencies.
According to comments by clients, suggestions and discussions related to
parenting appear to have primarily dealt with disciplinary issues and difficult child
behaviors. Several clients mentioned receiving assistance in learning other methods of
discipline aside from spanking. This assistance took the form of watching a video on
discipline for one client and then talking with her worker to learn how to put the
principles in practice. Other clients described discussing their discipline challenges with
their worker. One client explained that her worker had provided guidance and support
around discipline because she and her children "were having a hard time
communicating.. .[and] were really in a bad situation." Clients also described receiving
help managing their children's difficult behaviors. A grandmother caring for her teenage
grandson cited the challenges she faced in raising such an "active" child, and stated that
91
support from ARS will allow her to rest. This same grandmother also cares for her
granddaughter with Downs Syndrome and the ARS worker has helped her learn better
how to relate to and redirect the child: "I am finding a way [to discipline and redirect]
that suits her." A client who was having difficulty communicating with her children and
managing their behaviors said that she received guidance from her worker: "Now I know
how to calm my daughter down if she tantrums." The mother of a toddler said that her
worker spoke with her about strategies for weaning him off his bottle.
Sometimes parenting interventions emphasized the good things that parents are
doing. One client, a new mother, stated that her worker provided her with assurance that
her child was developmentally on track. Others also expressed appreciation for the
reassurances on their parenting provided by the ARS worker. This type of family
intervention is further discussed in the social support section.
Regular opportunities for families to enjoy their time together, supported and
encouraged by ARS staff, are a special service provided by the ARS-South Hayward
program. This program has teamed up with Lawrence Hall of Science, a local science-
focused children's center. ARS parents and their children are regularly transported to a
location in the community to participate in organized science, art, and other group
activities. These outings allow parents an opportunity to engage with their children in a
meaningful and often playful manner. Children and their parents work on projects
together and this experience is celebrated and honored by staff. The Hayward clinical
supervisor describes these activities as "a formula that is magic in some ways." The
experience of being together, and having their interactions and relationships honored,
makes families feel more competent and open to help, according to her description. A
92
number of clients also described their experiences at Lawrence Hall of Science as a
special time for themselves and their children. One mother said:
I really enjoyed the science classes.. .It was a nice experience and it taught me that the kids come first—they bug you and bug you, but they just want 5 minutes of your time.. .it was cool, I never had that when I was a kid. I'm always busy every day, then I have to go to work and clean, and only on the weekends is when I have time to do things with them.. .that day was cool, because I got to be with them.
Critical assessment of ARS parent-child relationship interventions
While the ARS program clearly recognizes the importance of the parent-child
bond, promotion of attachment may occur indirectly rather than directly. Direct
interventions related to parenting tend to focus on practical approaches to address
distiplinary or behavioral problems, rather than encouraging parental sensitivity that
might lead to better attachment. ARS staff described helping families deal with specific
issues related to their parenting, but did not emphasize training parents to pick up on cues
from small children or encouraging them to engage in developmentally appropriate play
activities. Only one client, a new mother, mentioned that her worker discussed
attachment and playing with her baby. This lack of direct attention to parent-child
attachment may be due to concern over lack of qualifications, as suggested by ARS-East
Oakland workers, or family crises that trump a focus on parent-child bonding, as
suggested by ARS-West Oakland. ARS-South Hayward appears to have the strongest
focus on the parent-child relationship, but it is unclear whether staff frequently make
interventions related to attachment when the opportunity arises or generally limit their
parenting-related interventions to addressing presenting problems in the parent-child
relationship, like challenges dealing with child behavior.
93
ARS provides services that may indirectly strengthen parent-child attachment. By
helping families meet their basic needs and connect to community resources, the ARS
program may help families decrease stress and provide parents with more time and
emotional resources for their children. Tickets to activities or structured activities like
the Lawrence Hall of Science program may provide families with time and space to enjoy
each others' company. In these ways, the ARS program may improve the emotional
relationships between children and parents. Attachment theory suggests that social
support may also reduce parental stress and give parents the resources to better connect
with their children. This intervention type will be discussed next.
Social support
Following distinctions made in the social support literature, ARS interventions
involving social support will be separately categorized and discussed as related to
perceived and enacted social support.
Perceived social support
Staff described the relationship they form with clients as one of the most important
aspects of the ARS intervention. They explain that many clients have no one else to
provide them with support. According to staff, the ARS program can give families the
feeling of being supported, albeit temporarily. A clinical supervisor emphasized this
point:
So many people we see feel isolated, whether because of culture, language, or feeling depressed. While they are in the program, there is the joining of home visitor and family, and the experience of being understood and accepted. Families are.. .left with a feeling of being supported. It is a quick period of time, but that is what I think is most valuable.
94
Clients also talked about feeling supported—as though they had a friend, someone
with whom to talk. They consistently described their relationships with ARS workers as
positive. While clients acknowledged that staff are acting in a professional role, they
often described feeling that their worker is more like a friend or family member. A
statement by a mother was illustrative of this common sentiment: "I feel like I have a
new friend. I can tell her anything. She can help me." A couple of clients described
their relationship with the home visitor as therapeutic, with the worker somewhat like a
counselor. More than one parent described her worker as an "angel."
Emerging from clients' comments was a sense that the relationship with the worker
was one of trust and availability. Several clients commented that their worker would call
them back right away, and would never leave them hanging. A number of clients
remarked that they had trust that the worker would respond to requests for help. This
statement by a client was typical of the sentiments of many:
It helps me to go on with my day easier by just talking to her, and knowing that there's someone who is willing to help. I know that someone is just there.. .to give you help whenever you need.
A helpful and supportive attitude seemed most important to clients. One client stated:
"It's not the money or what they give that's important, it's the way they talk to you."
Clients described a number of personal traits possessed by their worker that aided in
relationship development. These included kindness, a non-judgmental attitude, interest in
children, organization, patience, and dependability. Another quality was the willingness
to go "above and beyond" the call of the job, which might mean anticipating a parents'
need for diapers or providing a level of emotional support that seemed greater than that of
a worker for a client.
95
Clients described the feelings which have resulted from their relationship with their
ARS worker. A common theme was a general feeling of reassurance and calm. One
mother described how the opportunity to sit and talk each week relieved her of burden.
Another said that her worker had brought joy into her life. Clients described only
positive repercussions of their involvement with ARS, and particularly of the relationship
with their worker. Having the opportunity to discuss problems with a supportive person
was described by many clients as a calming experience. For example, when asked how
her life had changed as a result of participation in ARS, one mother stated: "Simply
talking helped." Awareness that the ARS worker is available for support is sustaining,
according to some clients, such as one who commented that knowing her worker is a just
a call away is a source of "empowerment and strength."
The themes discussed by staff and clients related to the perception of support appear
congruent with the theoretical literature. Clients tended to emphasize supportive qualities
and characteristics rather than specific acts of support. Workers described making
themselves available as key to the ARS intervention, instead of the provision of particular
types of support. Some clients appear to have been open and ready for help when offered
ARS services, a necessary antecedent for accessing social support, according to theory.
For example, one client remarked: "They are here to work for kids, for orientation and
support. I needed that very much." Some clients remarked on similarities between
themselves and their worker that made them feel more bonded. These similarities may
have helped clients see their workers more as a friend than as a professional helper. In
this way, the paraprofessional status of the worker could have been a factor that promoted
engagement. One mother, for example, commented that it was easier to relate to her
96
worker because they were of a similar age, while another commented on the shared
country of origin as a factor which brought them together. It is possible that the presence
of ARS staff in families' lives acts as a buffer against stress during the period of the
intervention. Statements from clients hint at this effect. A number of clients said that
awareness of the ARS worker's availability was reassuring and calming.
Enacted social support
Most often, clients stated that they accepted services initially due to the offer of
enacted social support—usually informational or instrumental. Often this took the form
of referrals to services such as child care or provision of material goods to meet basic
needs. Some clients also acknowledged that they accepted services for the promise of
enacted emotional support, as with one mother who said that she agreed to participate in
ARS because she wanted someone with whom to talk.
The type of enacted support provided to families appears in line with what they
need and request. For example, a first-time mom describes how her ARS worker brought
her information on child development and parent-child bonding and attachment, as well
as information on free activities in the community. A panoply of referrals and
interventions were described, and all the families interviewed said that their workers had
at least attempted to meet all their described needs. From the description of staff and
clients, a picture of distinct types of enacted emotional, informational, and instrumental
support emerged.
Emotional
Beyond simply being supportive and available to clients, ARS workers appear to
engage in specific types of emotional support. These tended to follow the descriptions in
97
the social support literature. One type of support is giving parents the time and space to
"vent" about the stresses and problems of parenting. This can be part of the engagement
process, as ARS staff describe allowing parents to talk about their feelings related to the
CPS referral without judgment. It can also extend beyond the initial phases of the
relationship, with clients describing their appreciation of the opportunity to share their
feelings with someone who is kind and caring. One client explains: "With her, I can
vent and get my feelings out and she supports me."
Another type of emotional support is the provision of reassurance and
compliments to parents. This type of support is also sometimes known as "appraisal
support." One ARS staff member remarked that clients often lack this type of positive
feedback from their own extended families. Pointing out and praising positive behaviors
are one way that ARS builds clients' strengths into their interventions. One worker
described how she will comment on strengths when her clients seem most confused and
lost, while another said that he will give a small compliment at every visit. It is
particularly important, according to staff, to remind clients of the progress that they have
made.
Normalizing the experience of asking for help is another type of emotional
support described by staff. One staff member explained that families might not be able to
care for their children because they do not have adequate resources. In this regard, ARS
workers can inform them of resources and reassure them that it is acceptable to seek out
help. As one staff member tells her clients: "This is the help. You just didn't know
about it and it's okay." One staff member commented that their image as a source of help
and support distinguish them from common views of CPS, which people tend to think of
98
as simply removing children. Help may be more acceptable from ARS because staff
make it clear that their purpose is to support, not to investigate, families. A number of
clients said that receiving help from their ARS home visitor made them feel good because
the worker was kind and non-judgmental; this comment from a client was illustrative: "I
was impressed because she offered me help instead of asking me what went wrong and
telling me things I should have done."
Informational
From the descriptions of staff and clients, several types of informational support
can be identified as part of ARS interventions. One such type of informational support is
the provision of factual information, such as on community services or on topics of
interest to families such as child development. This indeed was one of the primary roles
that ARS workers described for themselves. They might gather listings of housing
rentals, track down information on a particular service, or find handouts on a topic such
as discipline, depending on the needs of their client. Clients often commented on their
worker's knowledge of resources and of child development, particularly in contrast to
their own lack of knowledge. One mother commented that what was most helpful about
the program was that her worker knew about community programs for young children
and she, as a new mother and a newcomer to the neighborhood, benefited from that
information. Several ARS staff expressed the view that knowledge of community
services is particularly important for newcomers and immigrants. By providing
information, staff attempt to connect clients to their communities.
ARS workers also help their clients make sense of information. In this way, they
act as "another set of ears for them, when they have meetings where they might not fully
99
understand everything... [they] help them interpret, break it down for them," as one line
worker put it. This kind of assistance may be provided to families during appointments
or meetings, when the worker can help families understand what is going on and ask
questions on their behalf. It may also be provided when families are struggling to
complete paperwork, and the ARS worker can help explain the forms. One ARS worker
explains that the staff have experience and knowledge of navigating systems that they
impart to clients. Workers may also educate their clients by correcting misinformation,
as with one example a worker gave of a mother who believed that she might jeopardize
her efforts at establishing legal residency by accessing services for her child.
Clients described a number of examples of information requested and received
from their workers. These included requests for referrals, such as child care; facts, such
as developmental milestones; and advice, such as how to discipline children without
yelling. A number of clients also mentioned that their worker provided valuable
information on processes such as how to request an Individualized Education Plan for a
child, or how to format a resume. Having the right information made a difference to
families. As one client put it, her ARS worker "[fed] me the right info to make me strong
and confident and aware of my abilities to support my family."
Instrumental
Offers of instrumental support were frequently described by both staff and clients
as an initial hook for engagement in services. An ARS clinical supervisor explained that
gift cards and other material goods can encourage clients to accept the program and bond
with their home visitor. The line between informational and instrumental support is
somewhat blurred, because clients are often provided with information on referrals to
100
meet their needs and expected to follow-up with support from ARS workers. Attention to
basic needs could also be categorized as instrumental support, as could connection to
institutional resources; see these sections for further discussion.
One type of instrumental support described by staff is encouraging parents to
follow-up on referrals and helping them with the process. One way that staff do this is to
break things down for clients and help them see the small steps they can take to meet
their goals. A client describes one such interaction:
When I tell her everything that's going on, she'll say 'ok, what do you need to do first?' and if I say 'I don't know,' she says, 'c'mon Sarah (said with encouraging tone).' And then she'll help me if I need it."
Critical assessment of ARS social support interventions
The types of enacted support provided by ARS workers resemble those described
in the social support literature. From the comments of staff, ARS appears to fall within a
stress-buffering orientation. That is, specific acts of emotional, informational and
instrumental support are expected to reduce feelings of stress and lead to the development
of coping behaviors. While enacted support can provide particular instances of assistance
when needed by families, in themselves they are not likely to make clients feel supported.
This sense of perceived support appears to be created by the manner of ARS workers—
their qualities such as warmth and kindness and their actions such as returning calls
promptly and showing concern for families.
A question that arises from the remarks of staff and clients is how the feeling of social
support will be sustained after the intervention. Social support from home visitors is
intended to help families change attitudes and behaviors, not as an end in itself (Ekenrode
& Hamilton, 2000). With ARS, however, sustainable sources of social support appear
* Client's name changed
101
lacking. The staff themselves are the primary means of support, and therefore the source
of support is withdrawn at the termination of the intervention. Some clients commented
with regret about the anticipated end of the relationship with their worker after nine
months. However, some also acknowledged that termination was necessary, so that other
families might benefit from the program. The dependence on the staff for social support
is somewhat worrying, particularly for those who describe otherwise being isolated, such
as one client who commented: "I used to be depressed and so alone. Now I can call
[ARS worker]. Now I don't worry as much." Who, then, will provide support such
mothers after the nine month ARS intervention?
Research has identified preference for informal social support, but ARS does not
focus on making informal connections. A number of ARS staff remarked that they
encourage families to use support if they have it, but acknowledge that their clients'
extended family is often struggling. Because informal support is less available, that leads
staff to rely on established organizations. The one example that workers gave of utilizing
informal support was as a mental health intervention. For a grandmother experiencing
depression, her worker recommended that she regularly attend her church to seek support
from her fellow parishioners. Lawrence Hall of Science activities, conducted by ARS-
South Hayward, provide one opportunity for families to form relationships with each
other. However, ARS workers are not entirely encouraging of such friendships.
Relationships between their clients can be problematic, if families compare what they get
from the ARS workers. While ARS workers acknowledged that clients may have
difficulty establishing and sustaining friendships due to poor boundaries and other poor
social skills, they do not describe making specific interventions in this area.
102
The notion of reciprocity is embedded in the concept of social support. Families may
feel the need to reciprocate the help that they received. One mother described how she
enjoyed the chance to meet other families and build community at an ARS barbeque. She
said that the event made her feel that she might be able to make a positive difference in
the lives of others as the ARS worker had with her. Another client also said that her
experiences with ARS made her wonder how she might make a difference for others. A
third client remarked that her experiences in ARS made her want to help others,
particularly other women with similar life circumstances. As an improvement to the
program, a different client suggested that ARS make use of mothers like her who have
completed services to outreach to other families. Comments such as these suggest that
clients might appreciate a way to reciprocate the help they received from ARS by helping
other families. One staff member commented that the best way that a program like ARS
can help a community is to build self-sufficiency in families and position them so that
they are able to teach others. This could indeed be a deliberate focus of the program, to
encourage those families who are able and willing to give something back to their
communities. In this way, ARS could have a great transformative effect on
neighborhoods.
While the provision of support from the ARS worker lasts only nine months, ARS
workers put their faith in community resources as an ongoing source of support. Access
to services may not provide the warm personal relationship that families get with ARS,
but they will at least give parents specific types of services that they need, according to
staff. The role of institutional resources in the ARS intervention will now be considered.
103
Connection to institutional resources
In addition to the sense of being supported, ARS staff described connecting their
clients to community resources as the most important aspect of their work. They
discussed giving their clients a sense of what resources exist and introducing clients to
the process of using those resources. That might mean, for a family in South Hayward,
going with an ARS worker to the family resource center and meeting the family
advocates there. The purpose of connecting families to institutional resources, according
to staff, is to ensure that families have knowledge of where they can turn when in need of
help as well as skills in accessing services. As one of the clinical supervisors put it:
"There are some families that seem to be able, higher functioning to begin with, if they
can get the skills to be able to use the resources and you can plant certain things in there
that they can use, then nine months is fine."
According to ARS staff, they invest considerable time in locating and learning about
available resources. They describe building their knowledge of resources systematically,
often when first hired and learning about available services through colleagues or visits in
the community, and also case-by-case, as their store of knowledge grows with each
referral they make. One ARS worker referred to the "database in [her] head" that is
activated when she talks to clients and they describe their needs. Staff may learn about
services through networking with colleagues at other agencies during training sessions or
meetings. Another way they learn of services is through internet searches and the
Alameda County "Blue Book" service listing.
Participation in social services collaboratives appears to be particularly helpful for
identifying resources and connecting families. ARS-South Hayward has long been a
member of the South Hayward Collaborative, a well-established association of nonprofit
agencies committed to linking community resources through capacity building and
service integration. Staff of this program repeatedly highlighted the benefits of
participation. One benefit that was mentioned is having established contacts at other
agencies and at schools, which eases the process of referrals and cuts down on the time
that staff must spend navigating systems. One worker explained: "Just knowing or
having a contact person.. .you can give to your client, and call them... It opens a lot of
doors." Another benefit that staff described is the ability to connect families to a resource
center which remains as a place where they can return for assistance after the completion
of ARS services. Staff commented that they bring families to the local family resource
center "So when we're not over there not visiting them anymore they know 'Oh, we go
over there.'" Another benefit of the family resource center is the provision of a number
of services and material goods, such as Medi-Cal registration, immunizations for
children, and free clothing. Accessing services such as these at a 'one-stop shop' is an
efficient way of helping clients, rather than shepherding them around to different county
offices, according to staff.
Efforts to establish a collaborative have recently begun in West Oakland, with the
participation of ARS-West Oakland. While the association is new and not as established
as South Hayward's, staff are excited and hopeful about the new opportunities it has
provided to learn about local services and network with other providers.
The service context in East Oakland differs markedly. Many public agencies are
concentrated on the site of an old shopping mall that closed when it was no longer
economically viable in the community. This "service mall" offers a centrally-located
resource for neighborhood residents in need; however, the sheer size of the building and
its institutional image convey a different message from the community living room
suggested by a neighborhood-based family resource center such as that in South
Hayward. In addition to the mall there are a large number of smaller nonprofit and faith-
based services and programs, but these are not coordinated in any fashion and they are
not necessarily widely visible to community residents. ARS staff does not have
designated liaisons at the public agencies, nor do they have regular contact with the
smaller service providers in the community. As a result, when ARS staff works with
parents, they must determine the availability and accessibility of services, along with the
eligibility requirements for parents, slowing their work considerably.
Helping families acquire resources to meet their current needs is only half the task.
As staff describe it, their real goal is to implant skills so that families will know how to
meet their own needs after the ARS intervention is complete. This requires a fine balance
of helping families while not creating dependency. One staff member acknowledged this
by stating: "Myself, I don't do everything for them.. .1 let them do it, because we are for
9 months in their lives, and if they don't learn how to do it, then they will go back to how
they were when we entered their lives." Several workers emphasized the need to teach
families positive habits, and then wean them off assistance.
Institutional resources referral types
According to the comments of staff, the types of resources that families need varies
depending on children's developmental phases and the presence of any family issues. A
clinical supervisor pointed out that chronic issues, such as mental illness and substance
abuse, occur in a significant number of cases. Staff take a customized approach to
providing referrals, based on assessment of families' needs.
According to staff, when families first describe their needs, they tend to stick to
those which are concrete. Families may mention the need for a mattress for a child, or
assistance in acquiring health insurance. Willingness to engage is often related to the
family's expectation of realizing material gain. Rarely will families ask for help with
parenting at the beginning, according to ARS workers. This may come in time—as
families establish trust, they may become more open with expressing their parenting
concerns. As one ARS worker put it: "If we can show them that we come to the table
with something to offer, they're more willing to talk about other stuff."
The referrals described by staff and clients tended to fall within the categories
described in the institutional resources literature as important for children's development
and well-being, with a few notable additions. Housing was a particular concern because
of the lack of affordable housing in the Bay Area. Often families reside in sub-standard
or overcrowded housing, lacking other choices. ARS workers stated that they frequently
intervene with landlords to request maintenance work and address other problems, since a
request from a third-party can be more effective than those from the family alone. Legal
services were also a frequently-mentioned need, particularly for assistance with custody
battles, incidents of domestic violence, or efforts to establish legal residency. Access to
food was a concern for a number of families, to which ARS workers respond by
providing donations of food from the basic needs fund, referrals to food pantries, and
assistance in signing up for Food Stamps. Another frequent request is for assistance with
cash support programs, such as Social Security Insurance payments for parents and
107
children with disabilities or Temporary Assistance for Needy Families. Part of the reason
that these are frequently needed referrals may have to do with the ARS focus on the
needs of parents as well as children. Institutional resources theory describes services and
institutions that are generally important to children's development and well-being, not
necessarily those that address the needs of parents.
Aside from the previously mentioned referrals, other types of referrals tended to fall
within the main categories described by institutional resources theory: learning, social,
and recreational activities; child care; schools; medical facilities; and employment
opportunities (Levanthal & Brooks-Gunn, 2000). Learning activities referrals described
by ARS staff and clients included tutoring services and after school programs. The ARS-
South Hayward Lawrence Hall of Science program was described by several clients as a
beneficial learning opportunity for their children. One client mentioned that the program
helped to prepare her child for school. According to workers in West Oakland, parents
often ask for opportunities to spend fun, recreational time with their children, such as
tickets to the zoo or the movies. Organized recreational activities such as mommy-and-
me classes are another type of referral mentioned by clients. A new mother commented
that she enjoyed attending the mommy-and-me activities, free library story times, and
other community activities that her ARS worker told her about. These activities provided
her with valuable opportunities to enjoy time with her child, get out of the house, and
meet other mothers. ARS workers also provide referrals and sometimes funding for
summer recreational programs, to provide care and activities to children while they are
out of school. Clients also mentioned referrals to programs like swimming lessons and
computer classes.
108
Child care is an important need that ARS helps families address. Families are often
unaware that they may qualify for subsidies, according to staff. One worker remarked
that none of her clients whose children were of Head Start age had enrolled them.
Workers also mentioned helping families enroll their children in preschool in order to
promote school readiness. However, even when families are eligible for government
subsidies, it can be a challenge to actually get childcare, according to staff. Staff refer
families to one of the three child care resource and referral agencies in the county, but
stated that subsidized child care is frequently full, with long waiting lists. A number of
clients commented that their worker had helped them complete paperwork for Head Start
or regular child care, but that they were still on a waiting list. Three clients mentioned
receiving respite care. In at least one case, this respite care came from the ARS parent
agency (FSSBA).
Schools are an important locus of activity for ARS. Staff and clients mentioned
school-related referrals and activities including: help registering children for school,
advocacy with Individualized Education Plans (IEPs), and discussions with teachers
regarding children with poor academic performance. One mother stated that her son had
been struggling in school, but after the ARS worker helped to arrange an IEP and transfer
to a special education school, her son now loves attending school. At one agency, staff
mentioned that they try to make at least one visit to the schools and request children's
records to see if any problems can be identified. Staff may accompany parents on parent-
teacher meetings or other school-related meetings, where they can help parents by
interpreting information for them and advocating for their rights. As one worker pointed
out: "Oakland school district is difficult to navigate.. .Imagine if you don't speak English
or if you're non-native." Another way that ARS addresses school-related issues is by
screening for developmental delays. ARS staff conduct developmental assessments for
children and help parents access prevention and treatment services through schools if the
children are over the age of five, or Regional Center if children are ages 0-4. For the
ARS-West Oakland program, which since its inception has served children ages birth to
18, helping teens apply for college and financial aid is an important focus. If college is
not the teenager's plan, ARS staff stated that they will help their clients plan for their
future by telling them about Job Corp and other employment or training opportunities.
ARS-West Oakland staff stated that they frequently help teens mediate their relationships
with schools, particularly in cases of conflict or when a teen does not want to attend
school.
A number of different types of assistance with health and mental health needs were
mentioned by ARS staff and clients. According to staff, health insurance is a significant
need among their clients. Families often request assistance with applying for Medi-Cal
or other subsidized health insurance programs. An ARS worker told an anecdote of how
offering a family help with health care access encouraged them to engage in the program.
The family was initially reluctant to accept services, but the father shared that he had
been frustrated in his attempt to acquire insurance for their newborn daughter. The ARS
worker assured the family that she could help with the forms. Staff at ARS-South
Hayward mentioned that they are able to sign their clients up for Medi-Cal at the South
Hayward Collaborative's Family Resource Center, which makes their job easier. One
client mentioned how much she valued her worker's help in such matters:
110
The program is very important. They stopped my Medi-Cal services and we needed daily medication. [ARS worker] helped with setting up Medi-Cal again. It was very hard for me and he helped me a lot.
Clients also mentioned instances when the ARS workers helped with getting children
immunized or filling a prescription. One client mentioned a referral to an asthma
education program, so that she and her child could learn how to manage his health needs.
Therapy was a need often mentioned by staff and clients. Several workers
commented that their clients often experience depression, which may be identified by the
assessments completed by staff or later in the case. A large number of clients mentioned
that they, their partners, and their children are receiving therapy. Quite a few clients
reported that therapy is provided in the home. Clients reported that they and their
families had benefited from therapy. One mother told of how she and her son had
experienced domestic violence and the resultant feelings of nervousness and anxiety,
which therapy had helped to alleviate.
Employment is one of the three biggest needs for families, according to ARS staff
(along with housing and child care). Helping parents get a job may involve referring
them to a career center or job training program, helping to put together a resume, and
bringing job listings to home visits. One client expressed her appreciation for her
worker's assistance: "She's been a huge help with job search, like use of the computer,
things that I don't have access to."
The connection between ARS referrals and subsequent child development is not
systematically investigated, either by the ARS program or by this research project. It is
therefore unclear if, by making referrals in these areas, children's outcomes are
substantially improved. Nevertheless, the focus on addressing the types of institutional
I l l
resources identified as most important in the theoretical literature hints at the potential of
ARS to positively impact children's emotional, social, and cognitive development. The
qualities of institutional resources—their availability, accessibility, affordability, and
quality—also affect their influence on children (Levanthal & Brooks-Gunn, 2000).
Discussion will now turn to how ARS referrals reflect these dimensions.
Institutional resources dimensions
Availability
Availability of services varies markedly by neighborhood, according to statements of
ARS staff. Staff at ARS-South Hayward commented that services are less available in
their city than in Oakland, with its greater concentration of social services agencies and
its proximity to Berkeley, another service-rich area. The ARS-South Hayward clinical
supervisor remarked that there are clearly gaps in services in South Hayward, and that
available services often fill up. Staff commented that the lack of services in South
Hayward may make families more receptive to ARS, for the unique resources they can
provide, like participation in the Lawrence Hall of Science program. However, ARS-
South Hayward staff did acknowledge that those services which exist in South Hayward
are well-known to them and well-coordinated through the South Hayward Collaborative.
Likewise, in West Oakland, the consensus among staff was that resources are limited
in the area. Clients must often leave the neighborhood and go downtown to get services.
The staff stated that their lack of knowledge about resources was due in part to their
relatively new status as a program and because of the here-to-fore lack of networking
between ARS and other service agencies. Staff hope that this will change because of the
112
agency's involvement with the West Oakland Collaborative, which appears to offer
opportunities to meet representatives of other agencies.
By contrast, staff in East Oakland expressed their opinion that the services that their
clients need are generally plentiful in their neighborhood. One staff member related the
difficulties of trying to help a client who relocated to Sacramento identify services
compared to the task of finding comparable services in East Oakland:
I know here in East Oakland there's a lot of community-based organizations that are available for our clients. And you know, in addition to like the health services and mental health services and even daycare and anything. And I had a really hard time finding some of that stuff in Sacramento and I'm sure in other cities as well, you know.
While services are more available in East Oakland, ARS-East Oakland does not appear
well-connected to other agencies. Staff commented on their lack of connections and
liaisons with other agencies, which makes it more difficult to know about the availability
of resources.
Different types of services appear to be more or less available than others. At all
three agencies, staff commented on the difficulty of getting housing assistance for their
clients. Staff in Hayward commented on the lack of rental assistance programs. In East
Oakland, staff described the difficulties of getting public housing. West Oakland staff
described the challenges of helping families look for alternative housing when they can
not afford market rents. Employment is another area where staff struggle to help
families. While they can provide rental listings, assistance in putting together a resume,
or referrals to career counseling centers, staff in West Oakland point out that they do not
have any real connections to employers and are not always successful in helping their
clients gain employment. Child care can be tough to access. Staff in South Hayward and
West Oakland commented that childcare in their areas often have lengthy waitlists. Staff
113
in South Hayward described having difficulty getting therapy for their clients, which was
borne out by the experiences of clients, a couple of whom commented that they remained
on waitlists for therapy.
Availability of services also appears related to certain populations and their needs.
Services for mono-lingual Spanish speaking clients can be difficult to get, particularly
therapy. Workers lamented that resources are difficult in most cases to acquire for
undocumented families. The only subsidized service for which these families qualify is
Head Start. Aside from emergency medical treatment, undocumented families can not
get health care. A limited number of agencies will provide services to undocumented
families, but these services generally have long waiting lists, according to staff.
Accessibility
It is not enough to know that services are available—it is important as well that
families can access services. One mother shared that she now knows of services, because
of her participation in ARS, but does not have transportation to get to services and
consequently "feels stuck." ARS staff acknowledge that transportation is an issue, and
therefore make an effort to identify and refer clients to local services. Staff in West
Oakland commented that they make a particular effort to locate convenient child care,
and will occasionally provide families with bus passes. Transportation was
acknowledged to be a challenge in South Hayward by both staff and clients. The clinical
supervisor in South Hayward stated that bus service is limited. Among clients, those in
South Hayward most frequently described transportation as a challenge and a barrier to
access services.
114
Services provided in a one-stop shop setting can be most convenient for families,
allowing them to meet several needs in one trip. The Family Resource Center in South
Hayward plays this role, according to ARS-South Hayward staff. Clients can apply for
subsidized health insurance, get basic medical care like immunizations or medical tests,
receive clothing and food donations, and learn about other available resources, all at one
place. In East Oakland, a number of services, such as Food Stamps and Temporary
Assistance to Needy Families, are co-located together in a former shopping mall
converted to office space. This resource is a boon to the community, according to staff:
"Eastmont Mall, which I think for that community is very good because it's.. .easily
accessible by bus and also it's kind of a one-stop place that has a lot of services and it
has, its name is known."
Affordability
Staff describe the majority of families served by ARS as poor and struggling
financially to cover basics such as food and rent. The ability to pay for services is
beyond the ability of many, particularly for costly things such as child care. Staff
mentioned emphasizing to new clients that ARS services are free, as one selling point for
the program. ARS staff also seek out free and subsidized services for their clients
whenever possible. Most of the services to which clients are referred appear to be free or
low-cost. For example, staff in South Hayward report trying to access free therapy for
their clients by helping them enroll in Medi-Cal, and then making use of EPSTD (Early
Prevention, Screening, Treatment, and Diagnosis) funds. Housing and child care are two
resources that generally require some level of payment, which can be difficult for clients.
115
One worker remarked that even a sliding scale is too expensive for many of their families
when it comes to child care.
Quality & Competition
ARS staff said little of the quality of available services. The single relevant
observation was made by an ARS-West Oakland staff, who stated that quality mental
health services can be difficult to acquire for families because subsidized therapy is often
provided by student interns. Their frequent turnover and lack of experience can result in
therapy that is inconsistent and poor quality.
Competition for services was mentioned by staff with regards to demands placed on
scarce resources. Lengthy waitlists for some services, such as childcare, therapy, and
housing appears to indicate their scarcity and high demand. When scarce resources are
made available, such as public housing, word often travels fast. One staff member
related a recent conversation with a housing authority staff member. When she asked
about how housing lists were advertised, the staff member replied that advertising was
unnecessary—getting the information out by word of mouth was sufficient to draw large
numbers of applicants.
Critical assessment of ARS institutional resources interventions
Staff acknowledge that the ARS intervention is time-limited, and that resources must
be available in the community after the intervention to support families. This is their
rationale for teaching clients about the existence of community resources and skills of
how to access services. Staff attempt to plant seeds of positive behaviors and new skills
that with luck can be "just a brief exposure that may last for the rest of [their lives]." By
emphasizing knowledge and skills development for accessing resources, the ARS
116
institutional resources intervention may have longer-lasting effects on families' lives than
their social support interventions. This particularly may be the case since ARS focuses
on connecting families to the types of institutional resources that are theorized to be most
beneficial to children's development.
Participation on social services collaboratives appears to be highly beneficial to
programs like ARS that focus on connecting families with local resources. The South
Hayward Collaborative enhances services for ARS-South Hayward in two notable ways.
First, the knowledge and networks that staff have built through their participation ease the
process of referring their clients for services. Second, the Collaborative sponsors a
family resource center that provides a number of services to families on a walk-in basis.
After the ARS intervention, families in South Hayward will know where to turn in the
future when in need of help. This can make the gains these families experience while in
ARS more sustainable for the long-term. The West Oakland collaborative does not yet
offer the same level of knowledge and networking, nor does it have a drop-in center for
families, though these benefits may be realized over time. Families in East Oakland
appear at a disadvantage for the lack of a social service collaborative, though services are
generally more plentiful in their area.
Institutional resources theory suggests that services must be high quality, affordable,
available, and accessible to positively influence child development and parenting. ARS,
and differential response programs in general, do little to affect the service arrays to
which they refer clients. Because ARS is "a resource more than a direct agency.. .[the]
service is to provide resources and linkages," as one staff member put it, the program
might better serve clients by attempting to change and improve the array of available
117
services. The three community-based ARS agencies could do this by nurturing their
relationships with other services agencies and helping to foster fledgling social services
collaboratives like the one in West Oakland. Social Services of Alameda County, which
commands greater power and funding, could play a role in supporting the development of
new agencies. SSA has a vested interest in doing so, to enable families to receive
services within their community so that they do not become involved with costly
mandated child welfare services. By taking a leadership role to improve services across
these dimensions, ARS could partner with the communities they serve, rather than simply
putting new demands on existing resources.
Based on the comments of staff and administrators, there is reason to believe that
availability of institutional resources differs by neighborhood. Because connection to
institutional resources is one of the primary, and perhaps most important and long-lasting,
of the ARS interventions, variation in availability of resources may influence the
effectiveness of the program. The next chapter will examine the question of institutional
resources availability in the three ARS target neighborhoods, using Geographic
Information Systems software to organize social services data and analyze spatial
patterns.
118
CHAPTER 6: RESULTS (QUESTION #2)
Why examine social service arrays?
The success of alternative/differential response relies upon the availability of
neighborhood services to which families will be linked. From qualitative interviews with
ARS staff described in the last chapter, it is clear that the unique array of resources in
each neighborhood affects program implementation. ARS staff described prioritizing
referrals within families' neighborhood of residence because many lack cars and may not
follow-up on referrals out of the area.
This portion of the research seeks to understand the connection between resource
distribution and ARS program implementation. The research question to be answered is:
What is the resource distribution in Alameda County and how might institutional
resources in neighborhoods influence ARS implementation? Analysis was conducted at
two levels: zip code level, because ARS services are targeted by zip code; and census
tract level, because this geographic designation may more closely resemble what people
consider to be their neighborhoods. The eight zip codes served by ARS were prioritized
in the analysis, as were the census tracts that make up the zip codes (see Figure 6 for a
map of Alameda County, with ARS zip codes marked).
Two main concepts were considered by this portion of the research: need for and
availability of social services. The concept of need for social services was
operationalized as rates of child maltreatment reports. Need was interpreted in this way
because ARS services are targeted to neighborhoods (designated by zip code) with the
highest rates of child maltreatment reports. Availability was defined as total number of
services in an area, taking into account the total area of the given zip code or census tract.
119
A one mile buffer was created around each zip code and census tract, and services located
within this buffer were included in the total number of services available to residents in
the given zip code or census tract. The buffer was added out of an understanding that
administrative boundaries are invisible and easily traversed. Services within the area
nearby a zip code or census tract are also understood to be available to those residents.
The types of services included in the analysis are a combination of those services
identified as important for children's development by institutional resources theory
(Levanthal & Brooks-Gunn, 2000) and those services acknowledged by ARS staff as
most important to the families they serve. Data on need and availability were gathered
and used to create "layers" of aggregated spatial data.
Data collection
Need
Data on child maltreatment report rates was acquired from the California Child
Welfare Archive at the Center for Social Services Research, UC Berkeley. This archive
contains data on all child maltreatment reports made in the state of California at the
report-level, including demographic and address information for children reported as
possible victims of child maltreatment. The data received were the total number of
reports by zip code for 2004 and 2005, and by census tract for 2003 to 2005.
Availability
Data on social services locations were acquired from a number of sources. The
data collected were the name of the agency, the service type, and the address (some
sources also included additional information not utilized in the analysis). The first and
most comprehensive source of data was Eden Information & Referral, a nonprofit agency
that collects data on social services and makes this information available to the general
public on a free website and a 'blue book' available for purchase. Data were pulled from
their records, for a fee, for certain service categories. Data on child care centers and
family child care providers were provided from the three Alameda County child care
resource and referral agencies. Alameda County agencies provided data on health and
mental health services. The final source of data was the internet, since some of the
desired types of data—namely, churches and Alcoholics Anonymous and Narcotics
Anonymous meeting sites—were only available on the internet. Data from the internet
were downloaded into text files. An AML program was used to clean the data, put it in a
standardized format, and output the data into a delimited file which could then be read
into ArcGIS 9.0. See Table 3 for a list of data types.
Data management
Need
Child maltreatment rates were averaged over the available time periods, and
normalized by child population (as incidence per 1,000 children). The values were then
divided equally into low, medium, and high categories. Zip codes targeted by the
Another Road to Safety program were highlighted on the maps, using the symbology
options, with a thick black border.
Availability
Once the necessary data on service locations and types were secured from the
sources described above, they were cleaned by removing duplicates. Data were next
geocoded using ESRI's Streetmaps USA as a reference file to acquire the geographic
coordinates. Addresses that did not match were checked to see if parts of the data were
incorrect, using Google Maps. In a number of cases, the zip code was incorrect m the
data file and substituting the correct one from Google Maps resulted in a match.
Availability was calculated as location within census tract or zip code and a one
mile buffer. To calculate these totals, the geocoded service locations data were joined to
a zip code file and a census tract file for Alameda County, using the option "falls
completely within polygon." Using the Arc function of ArcGIS, coverages were built to
determine the areas of each census tract and zip code. A coverage is a data model that
stores related geographic features and automatically calculates the area of each polygon
(in this case, census tract or zip code) in the building process (ESRI, 2006). An
automated program was developed to create a 1 mile buffer around the zip and tract
polygons (see Figure 7 for a picture of the zip codes and census tracts with buffers).
Number of services within the buffer and polygon for each zip code and census tract was
calculated, and normalized by area.
Data analysis & findings
Zip codes: Need for & availability of services
Descriptive statistics were calculated for each zip code as well as its respective
buffer. The average number of services per ARS zip code (1415) was substantially
higher than the average throughout the county (1027). Extending out to include the
buffers, the average number of services available to residents of ARS target zip codes
(1681) was again much higher than the county average (1172). ARS zip codes with their
buffers differ by a margin of 492 available services, a large number, though smaller than
the standard deviation for the county as a whole (551 for zip codes with buffers). See
table 4 for numbers.
122
Spatial patterns were next examined for need and availability of services (see
Figure 8). All zip codes in West Oakland and East Oakland were classified among the
1/3 of zip codes with the highest rates of child maltreatment reports. Of the South
Hayward zip codes, two were classified as among the highest incidence and one as
among the middle incidence for the county. These patterns reflect the original reason for
selection of ARS target zip codes, as those neighborhoods with the highest rates of
maltreatment reporting and thus need for secondary prevention services. With regards to
service availability, West Oakland was categorized as among the 1/3 of zip codes with
the highest number of services, which contradicts statements by staff in West Oakland
that few services are available in their area. East Oakland has a mix of high and medium
availability zip codes, supporting the statements by staff that resources are fairly
numerous in their area. Likewise, South Hayward has a mix of medium and low
availability zip codes, which is consonant with statements by ARS-South Hayward staff
that fewer services are available in their area as compared to Oakland. To get a sense of
the relationship between service availability and child maltreatment report rates,
correlations were run in SPSS. There were no statistically significant relationships in the
case of zip codes (p-value = 0.393).
Census tracts: Need for & availability of services
Viewing need and availability of social services data at the census tract level reveals
pockets masked at a more aggregated level (see Figure 9). Census tracts did not perfectly
fall within ARS zip codes, and for this reason descriptive statistics related to the ARS
census tracts were not run. Looking at census tracts bounded by zip codes did reveal
useful information on distribution of child maltreatment rates and social services density.
123
Zip code areas with high rates of child maltreatment reports are composed of census
tracts that have a mix between high, medium, and even low incidence. West Oakland has
a high incidence of child maltreatment report rates at the zip code level, but by census
tract, about half the area is medium incidence and a small corner is low incidence.
Similar patterns are evident for high incidence East Oakland and South Hayward zip
codes.
Patterns of social services distribution also appear different at the census tract level.
The West Oakland zip code is made up of census tracts that are mostly low availability,
with higher availability census tracts concentrated in the areas close to downtown
Oakland. This pattern corresponds with ARS-West Oakland staff descriptions. In East
Oakland, services are centrally located in or near Eastmont Mall or near the freeways.
ARS-East Oakland staff had commented that many public services were clustered in the
Eastmont Mall, a one-stop shop for public social services. Mostly characterized by
census tracts with low availability, Hayward appears to have fewer services than the
northern county and the Berkeley/Oakland area, which also reflects the comments from
South Hayward staff that fewer services are available in their community than those of
the other ARS program sites.
There was a statistically significant correlation (Pearson's Correlation=0.145)
between child maltreatment report rates and availability of services at the census tract
level (p=.009). There appears to be more social services in areas with higher incidence of
child maltreatment reports. While incidence and availability were normalized to account
for population and area, this relationship may be related to urban density. There may be
more child maltreatment reports as well as social services in urban areas, so the observed
relationship between child maltreatment reports and density of social services may be
confounded.
Conclusion
Spatial data patterns reflected the resource distribution described by ARS staff,
once data were considered at the census tract level. Since connection to institutional
resources is one of the main interventions provided by the program and many families
rely on local resources due to lack of transportation, differences in the availability of
services within the different ARS neighborhoods may significantly influence experiences
and outcomes associated with program participation.
These analyses have a number of limitations. Reports of child maltreatment was
used as a proxy for understanding potential need for services because case-level data on
client service need and usage could not be obtained. The cross-sectional design does not
directly test the influence of resources on child welfare outcomes; this would be best
tested with a prospective, longitudinal study. While there was a careful effort to gather
all available social services data for Alameda County, it is difficult to know how many
services relevant to families were actually captured by the analysis since a complete
accounting of services does not exist.
There are few empirical tests of institutional resources theory. Future research on
differential response could add to the literature on the quality, availability, affordability,
and accessibility of institutional resources by examining these dimensions with regard to
client outcomes. For the ARS program specificially, the next step with this research
would be to disaggregate client needs by neighborhood in order to identify the principle
needs of ARS clients and whether they can be addressed within their communities. For
example, if clients mostly need food or low-cost diapers, are there food banks and
grocery stores in the area? This line of research could help identify situations where
clients should be referred to other communities, or where county social services might
invest to build social services infrastructure.
CHAPTER 7: RESULTS (QUESTION #3)
Background on statistical analysis method and hypotheses
While the ARS program has many proximal goals—increased connections with
community resources, provision of temporary social support, elimination of unmet basic
needs, and improvement of parent-child relationships—the overarching individual and
systems-level goal is to reduce the likelihood that families will enter the child welfare
system. Differential response evaluations have typically focused on outcomes associated
with child welfare system involvement, with less attention to family changes in other
domains, and this portion of the study follows that convention. The research question to
be answered is: Is ARS successful in preventing future child welfare system
involvement? Future child welfare system involvement is operationalized as re-report,
investigated re-report, and substantiated re-report of child maltreatment following the
nine-month ARS intervention. Differences in time to these events between the treatment
and comparison groups are analyzed using the statistical method of survival analysis.
Survival analysis is a type of analysis used for data that conform to a structure
with a defined time origin and end-point. Time origins often represent participant
recruitment into a study, the beginning of participation in a treatment program, or
diagnosis with a medical condition. The end point is generally considered a "failure"
event and in medical research may represent death (hence the term "survival analysis").
Data of this kind are better handled by survival analysis than standard regression methods
for several reasons. First of all, survival times of similar individuals are typically
positively skewed. This pattern violates the ordinary least squares regression assumption
of normal distribution. Second, for a variety of reasons, end point data may be
unavailable. This phenomenon is known as censoring. The most common form, right
censoring, exists when an individual has not yet experienced the failure event when the
study ends, or the individual is lost to follow-up. Left censoring occurs when the actual
time at risk is less than the time observed; the origin occurs before a known failure date,
but the exact origin time is unknown. Data may also have interval censoring in cases
where measurements are taken at spaced intervals but the failure event takes place in
between measurements (Collett, 1994). Longitudinal data analysis methods like survival
analysis also do a better job of accounting for time-varying explanatory variables
(Allison, 1984). An example of this kind of data would be age or any other covariate that
is not fixed and may change over time. In this analysis, data are likely to be right-
censored because families may experience the failure events after the period of time
monitored (May 2002-December 2007). The covariate included in this model (prior
reports) may vary over time, but this variable is only measured once at the end of the
study period, so methods to account for time-varying variables are not used.
The hazard ratio is reported with its significance level for each type of failure
event (child maltreatment re-report, investigated re-report, and substantiated re-report).
This number represents the odds that a family, given treatment, will experience the
failure event. It compares the hazard rates of the treatment and comparison groups. The
hazard rate is the probability that the failure event, if it has not already occurred, will
occur in the next time interval, divided by the length of the time interval. The time
interval is made very short to provide a practically instantaneous rate. The Cox
Proportional Hazard Model used in this analysis assumes that the hazard ratio is constant
over time (Spruance, Reid, Grace & Samore, 2004).
128
The hypothesized effects of ARS treatment on subsequent child welfare system
involvement are somewhat complex due to the potential bias that may arise from
increased surveillance of families referred to the program. ARS clients may be more
likely to be re-reported than members of the comparison group because they are known to
the system and have ongoing contact with the community providers to whom they have
been referred by the ARS program. For this reason, treatment may not reduce re-report
and may in fact increase it (Chaffin et al , 2001). How treatment might affect
investigation is also somewhat of an unknown: since investigation is limited only to cases
in which children appear to be in danger or at significant risk, presumably fewer ARS
clients would be investigated. Then again, the same surveillance bias may again arise,
resulting in greater rates of investigation among families known to the system. The
anticipated effects are clearest in the case of substantiation, the finding that a report meets
the statutory definition of child maltreatment. The ARS program would be expected to
reduce this outcome. However, due to the small sample size and the rarity of its
occurrence, this study may have limited statistical power to address the outcome of
substantiation.
Data collection
Data were provided by Alameda County Social Services Agency from the Child
Welfare Services Case Management System (CWS/CMS) and from the Another Road to
Safety ECChange database. Prior to receiving the data, a research assistant (MSW
student Anna Geer) spent several days at the SSA office, assisting in data entry and
cleaning. This process involved cross-checking the CWS/CMS and ECChange databases
to confirm which families were referred to services, using identification numbers when
129
available or child names when they were not. A certain amount of error may have been
introduced at this step, as children with names similar to children referred to ARS may
have been misclassified. The first portion of data were provided on December 18, 2007;
this initial run included the demographic and child welfare histories for all families
whose files indicated that they had been referred to the ARS program, as well as families
who met the eligibility criteria of child age, zip code of residence, and evaluated out
report, but were not referred due to reasons of program capacity. A second data pull was
made from ECChange on January 25, 2008. These data provided more accurate
information on which referrals were received and processed by the ARS agencies and
which were retained for services. This file also included demographic information,
which was used in the analysis as it was more likely to be accurate than the data from the
screening reports contained by CWS/CMS. Upon examination, it became apparent that
this file contained some inaccuracies in the demographic information (for example,
gender clearly did not match the child's presumed gender based on name), so the data
were re-run on February 14, 2008, with only those individuals selected as part of the final
treatment group.
For several reasons, the decision was made to restrict data analysis to families
referred to the ARS-South Hayward program. Of the three ARS sites, this program is the
most well-established according to SSA and First 5 administrators. It has also existed for
the longest, as the host agency (La Familia) was one of the two agencies to participate in
the original request for proposals. Administrators at SSA and First 5 also expressed the
view that this site had maintained closest fidelity to the original model. For these
reasons, administrators felt that the most valuable lessons could be learned by examining
130
this site alone. Analyzing data from only one site also minimizes variation of treatment
that likely exists across program sites.
Data management
Data were provided in excel sheet format, and managed in this format and in Stata
".dta" format. The following steps were taken. First, the treatment and comparison
group samples were constructed out of those files provided. Next, data were formatted in
the manner required for Stata to conduct the survival analysis. Finally, the data were fit
with nonparametric and parametric models, to analyze the effect of treatment on the
outcomes of re-report, investigated re-report, and substantiated re-report nine months
after the index date (either date of referral to ARS or date of first evaluate out report
within the study period, for the comparison group). Failure events were assessed nine
months or later post-referral (the length of the ARS intervention) in order to examine the
full effects of treatment and minimize surveillance effects resulting from weekly ARS
home visits.
Sample construction for the treatment group relied upon the criteria of being
retained for services by ARS and including only one child per family in the analysis, to
preserve the independence of the sample. At the initial home visit, the home visitor and
clinical supervisor conduct an actuarial assessment and make a decision about offering
services to families: families may be referred to community services if they are assessed
as low risk; families may be retained for ARS services if they are assessed as moderate to
very high risk; or families may be returned to SSA if they refuse services or if the
decision is made that the families' safety or risk concerns merit action by child protective
services. The original ECChange data file was sorted by disposition, and only those
131
families who were retained for services with ARS-South Hayward were selected (n=265).
A portion of this sample had no CWS/CMS-assigned identifiers, meaning that it was not
possible to get the records of their child welfare system involvement. SSA staff was able
to locate identifiers for a portion of the cases, but 42 cases had to be dropped because
identifiers could not be located, bringing the sample to n=223. Next, duplicates were
dropped from the sample (10), bringing the total to n=213. Based on the common report
identifier for families, siblings were hand-coded as " 1 " if they would be retained for the
sample and "0" if they would be dropped. Siblings were dropped for several reasons.
One reason is that child abuse and neglect is a family problem; often multiple children in
the family are affected and included in maltreatment reports. Keeping siblings in the
sample would have violated the assumption of independence between cases. Another
reason is that ARS is a family-level (not child-level) intervention, since services are
provided to all family members, not just a target child. By retaining only one sibling, this
becomes a family-level analysis. Siblings were coded systematically, row by row: for the
first set of siblings, the youngest child was retained and the oldest deleted, with the next
set as vice-versa. Fifty-two children coded as siblings were dropped from the analysis,
putting the treatment sample at n= 161.
The comparison group was constructed with families who, according to Alameda
County Social Services Agency staff, would have been eligible for the ARS intervention
but were referred during a period of program capacity. Families met the same referral
criteria as ARS clients: screened out of investigation and residence in South Hayward
target zip codes. The original file provided by Alameda County SSA included data on
1634 ARS-eligible reports made between May 2002 and February 2007. Of these,
children residing in non-South Hayward zip codes were eliminated (n=687). A cross
check of the comparison group with the treatment sample revealed that 20 children had
been referred to the ARS-South Hayward program; these were shifted to the treatment
group. Siblings were again coded in the same manner described above; all but one
sibling from each family was excluded from the sample (n=208), resulting in a
comparison group of n=477.
Data for the treatment and comparison groups were next organized and hand-
coded. An initial date was selected: date of referral to ARS for the treatment group and
date of initial evaluated-out report for the comparison group. A column was created with
the date of referral plus nine months (approximated as 270 days), in order to determine
the failure date. The first failure date that fell after the nine month mark was selected as
the failure date. For clients with no subsequent re-report, the date which marked the end
of the study (February 14, 2007) was selected. Data for only the first report was included
for each client. In some cases, there was more than one re-report in a short period of
time. Following the convention of Alameda County SSA, two reports within a five day
span were considered to reflect the same incident of maltreatment. The most serious
agency response within the five day period was counted as the failure event, according to
the following hierarchy used by SSA: immediate investigation, 10 day investigation, or
no investigation/evaluated out. Data on the maltreatment report, client demographics,
and prior child welfare history were included in the final analysis file. Failure events
were coded as binary data (1 for yes, 0 for no) for re-report, investigation of re-report,
and substantiation of re-report. The following additional information was included for
each client with regards to the "failure" maltreatment report: allegation (physical abuse,
sexual abuse, emotional abuse, or neglect); report response (immediate investigation, 10
day investigation, or no investigation/evaluated out); and investigation conclusion (not
applicable for cases with no report or investigation, unfounded, inconclusive, and
substantiated). Demographic information (date of birth, ethnicity, and gender) were
included for each child. Total child maltreatment reports prior to the index date were
tallied and included as a binary variable (0 for none, 1 for one or more).
After sample construction and coding, MSW student Anna Geer conducted a
quality assurance check on the sample construction and data coding. She checked the
treatment and comparison group samples to ensure the following: all participants met
criteria for study inclusion, no participants were mistakenly excluded, index and failure
dates were accurate, and data coding for number of prior reports was correct. Some
errors were identified and as a result, 34 children were added to the comparison group
who were found to be excluded without reason. One participant in the treatment group
was identified as not meeting the criteria of residence in South Hayward and was dropped
from the sample. Information was replaced for 21 children in the comparison group and
7 children in the treatment group because either the failure date or number of prior
reports was incorrect. After corrections were made from the QA process, the final
sample was n=160 for the treatment group and n=511 for the comparison group, for a
total sample of n=671.
Data analysis & findings
Descriptive statistics
With one notable exception, the treatment and comparison groups did not differ
significantly demographically. Gender distribution nearly equivalent: 50% of the
134
treatment sample and 55% of the comparison group was male (Chi-Square 1.32, p =
0.251). Primary ethnicity was also fairly similar: 39% of comparison group and 28% of
treatment group were Hispanic (Chi-Square 6.91, p=0.009); 21% of the comparison
group and 12%) of the treatment group were Black (Chi-Square 7.06, p=0.008); 24%) of
the comparison group and 18%> of the treatment group were White (Chi-Square 2.17,
p=0.141); and 10%> of the comparison group and 16% of the treatment group were Other
(Chi-Square 4.52, p=0.033). The only significant difference with regard to ethnicity
(Chi-Square 50.57, p=0.000) between the samples was that ethnicity was more frequently
reported as unknown for the treatment group (27%) than the comparison group (6%),
which is surprising given that the families in the treatment group had more sustained
contact with workers. Better records on family demographics may be maintained by
ARS-La Familia in a separate database or paper files. At the time of ARS referral or
index report, the ages of children were largely equivalent: 28% of the comparison group
and 38% of the comparison group were infants (Chi-Square 6.17, p=0.013); 55%) of the
comparison group and 45%) of the treatment group were preschoolers (Chi-Square 4.88,
p=0.027); and 1% of both the comparison and treatment groups were elementary school
age (Chi-Square 0.0030, p=0.957).
The most significant difference between the treatment and comparison group is
contact with the child welfare system prior to ARS referral for the treatment group, or
index report for the comparison group. Ninety percent of the treatment group had one or
more prior child maltreatment reports (with 31% of these having two or more reports),
contrasted to 66%) of the comparison group (of whom 13% had two or more reports) with
Chi-Square 151.87, p=0.000. These numbers seem too dramatically different for mere
135
coincidence, suggesting that hotline screeners may have more frequently referred to ARS
clients with a history of prior reports. This would hardly be surprising, given that prior
re-report is a well-established risk factor for future reports (Marshall & English, 1999;
Baird, 1988; Baird, Wagner, & Neuenfeldt, 1993; Schuerman et al., 1994). Because re-
report is associated with referral to the ARS program as well as the outcomes of re-report,
investigated re-report, and substantiated re-report, it is likely to be a confounder in the
analysis. See Table 5 for the demographic data on the treatment and comparison groups.
The treatment and comparison groups were fairly similar in the experiences of
subsequent re-report, investigated re-report, and substantiated re-report, with a few
notable exceptions. In both cases, about 30% of the sample experienced a re-report
(n=48 for the treatment group and n=163 for the comparison group; Chi-Square 0.20,
p=0.652), suggesting that treatment was ineffective in preventing this outcome. Re-
report tended to occur faster among the treatment group: the ratio of the average time to
re-report for the treatment and comparison groups is 641:755 days. Neglect and physical
abuse were the most common re-report allegation types for both groups. Re-reports were
more likely to be investigated for the treatment group (n=38, 79%) than for the
comparison group (n=66, 40%), though this difference was not significant (Chi-Square
1.28, p=0.26). The types of investigation were, however, quite similar: 58% of the
treatment group got an immediate investigation, contrasted to 66% of the comparison
group. Investigation conclusions were also similar, with no significant difference in
substantiation (24% in the treatment group, 18% in the comparison group; Chi-Square
1.38, p=0.240). Neglect and sexual abuse were the most frequently substantiated
allegation for the treatment groups (both at 33% of all substantiated reports), while
neglect and physical abuse were most frequent for the comparison group (61% and 22%,
respectively). See Table 6 for a comparison of clients with re-reports, investigated re-
reports, and substantiated re-reports.
Risk scores from the Standardized Decision Making tool (SDM) were available for
ARS clients, though not for the comparison group, so therefore risk scores were not
included in the statistical models. ARS home visitors, together with their clinical
supervisor, complete the SDM at an initial meeting with the families. For the 160 clients
included in the treatment sample, 29 were missing risk scores. Of those with risk scores,
the majority fell within the moderate (40%) or high (33%) range, with small numbers in
the tails of low (2%) and very high (8%) risk.
Outcome: Re-report
Two nonparametric approaches, which made no assumptions regarding time to
event, were first used to examine the data. First, a log-rank test was conducted to see if
there was evidence that one of the groups was failing faster. The null hypothesis is that
the survivor functions of the two groups are the same. There was no evidence to reject
the null (p-value=0.82). Actual totals for the failure event tended to be slightly greater
than expected for the treatment group and slightly lower for the comparison group, but
the variation was not statistically significant. A Kaplan-Meier survivor function was also
plotted. The treatment group appeared to fail faster than the comparison group. Toward
the 1500 day mark following referral to treatment, the survival for the treatment group
begins to flatten out. The estimated hazard function shows that hazard starts out higher
for the treatment group, but after the first couple of years following referral to treatment,
137
the hazard for the treatment group falls to a level below that of the comparison group.
See Figure 10 for plots.
In the first wave of parametric analysis, a Cox regression was fitted with
treatment and other covariates. The initial model, with treatment alone, yielded a hazard
ratio of 1.04 (confidence intervals 0.75, 1.4) and p-value 0.83. There is only a very slight
trend of increased risk for the treatment group, and the confidence interval is fairly
evenly distributed around 1, suggesting no effect of treatment on re-report. Binary
variables representing the demographic information displayed in Table 5 were added
independently with treatment to fit several models: Male gender (HR 0.98, p-value 0.89);
Hispanic (HR 0.99, p-value 0.93); Black (HR 0.92, p-value 0.65); White (HR 1.4, p-value
0.025); Ethnicity-other (HR 0.92, p-value 0.72); Infant (HR 0.86, p-value 0.34); Toddler
(HR 1.02, p-value 0.93); Preschool (HR 1.13, p-value 0.365). Sample sizes for reports
with pregnant/newborn children and elementary school age children were too small to
allow for inclusion. The only significant covariate was prior reports, with a hazard ratio
of 1.78 (1.3, 2.4) and p-value 0.000; when adjusting for prior reports, the hazard ratio for
treatment dropped to 0.74 (0.52, 1.06) with p-value 0.104.
The proportion of children with prior reports is significantly higher in the
treatment group than in the comparison group, and number of prior reports may affect
both assignment to treatment and the outcome of re-report. A variable representing the
interaction between treatment and prior reports was created and added to the analysis
with treatment and prior reports variables. This model provides evidence of interaction.
Treatment effects differ depending on whether clients have prior reports. The effect of
treatment on re-report among those with one or more prior reports is HR 0.69 (0.48,
138
1.01), p-value=0.05; for those with no prior reports the effect of treatment is 1.46 (0.54,
3.98), p-value 0.74. Because there is little information on clients without prior reports
(n=16), the estimate of treatment effects for those with no prior reports shows poor
precision, with a wide confidence interval that spans one. There is a non-statistically
significant trend for treatment to prevent re-report among clients without prior reports.
Outcome: Investigated re-report
The next series of analyses examined whether treatment affected investigation.
As was mentioned previously, it is conceivable that the treatment would not reduce re-
report due to surveillance bias but would improve family functioning and the severity of a
child maltreatment incident and thus likelihood of investigation. For the entire sample,
138 families experienced an investigation; of these, 100 were in the comparison group
and 38 were in the treatment group. For the log-rank test, the p-value was not significant
(p=0.12) and therefore the null hypothesis of equality of survival functions cannot be
rejected; however, the log rank test suggests that fewer members of the comparison
group, but more members of the treatment group, experienced investigation than would
have been expected. The Kaplan-Meier survival curve showed that the treatment group
experienced investigation at a faster rate than the comparison group, and the hazard
function depicted a higher hazard of investigation for the treatment group until about
1250 days after referral to treatment (See Figure 11).
Fitting parametric models to the data with investigation as the outcome yielded
similar findings to re-report as outcome. For the unadjusted model, there was a trend for
treatment to increase risk of investigation: HR 1.34 (0.93, 1.96), p-value 0.119. Models
fit with demographic variables were not significant, with the exception once again of
139
prior child maltreatment reports. Adjusting for prior child welfare reports, the hazard
ratio associated with treatment was HR 0.96 (0.63, 1.47), p-value 0.87.
Rates of investigation were again found to differ depending on whether families
had a history of child maltreatment reports. For families with prior reports, the effect of
treatment is HR 0.89 (0.57, 1.39), p-value=0.63. With no prior reports, the effect of
treatment is HR 1.84 (0.58, 5.91), p-value 0.3.
After looking at the outcome of investigation with the full sample of clients who
received the treatment, another analysis was conducted to examine the rates of
investigation only among those with re-reports. The question for this portion of the
analysis was: Among those with re-reports, does treatment affect whether a family is
investigated or not? Families were dropped from the analysis if they did not experience
re-report. In this analysis, treatment appeared even more likely to be associated with
investigation. In the log-rank test, the margin from expected was wider and the p-value
was significant at the p<0.005 level (p=0.002), providing support to reject the null
hypothesis that the survivor functions of the two groups are the same. The Kaplan-Meier
curve for the treatment group fails dramatically faster, and the treatment group has an
elevated hazard until about 1400 days after treatment. For the unadjusted model,
treatment by itself is significant at the level p<0.005 (p=0.002), with a hazard ratio of
1.81 (1.24, 2.65). Adjusting for prior reports, the hazard ratio falls to 1.58 (1.04, 2.40)
with p=0.33. There was no evidence of interaction: treatment effects for those with prior
reports was HR 1.58 (1.00, 2.48) p= 0.045 and treatment effects for those with no prior
reports was HR 1.56 (0.49, 5.00), p=0.45.
For those families who experience a re-report, treatment appears to increase the hazard of
investigation.
Outcome; Substantiated re-report
Similar patterns were evident for substantiation as the previous outcomes, with
the main difference being that substantiation is a rare event (n=9 in treatment group;
n=l 8 in the comparison group), and so the power to detect differences between the two
groups is smaller. For the log rank test, the null could not be rejected (p=0.17), though
again values for the treatment group were higher than expected, and lower than expected
for the comparison group. The survival curve for the treatment group was slightly lower
than that of the comparison group, with concomitant higher hazard (Figure 12). The
initial unadjusted Cox regression model with treatment alone suggested an increased risk
associated with treatment, though the confidence intervals were wide: HR 1.75 (0.78,
3.89), p-value 0.17. None of the demographic variables were significant except prior
child maltreatment reports. When adjusted for prior reports, there was a 0.88 (0.37, 2.09)
hazard associated with treatment (p=0.78). There was evidence of interaction, with a
0.74 (0.30, 1.82) hazard associated with treatment for those with prior reports (p-
value=0.519) and a 5.47 (0.65, 45.70) hazard associated with treatment for those without
prior reports (p-value=0.117). Confidence intervals are again so wide, and the power of
the sample so weak, that no definitive conclusion can be drawn. See Figure 12 for
survival curves and hazard functions.
As with investigation, another analysis was conducted to examine the rates of
substantiation only among those families who were investigated. This portion of the
analysis sought to answer the question: Among those families who were investigated,
141
does treatment affect substantiation? The log-rank test was not significant (p=0.23),
providing no support to reject the null that the survivor functions are equivalent. The
survival estimates depict a slightly lower rate for the treatment group, and a higher degree
of hazard. In the unadjusted model, there was a 1.63 (0.72, 3.67) hazard ratio associated
with treatment (p=0.239); as with the previous analysis, the confidence interval was wide
and spanned one, making it difficult to draw a conclusion about the effect of treatment.
Adjusting for prior reports, the hazard associated with treatment dropped to 0.89 (0.37,
2.16), p-value=0.80. Another model was developed, with an interaction term for
treatment and prior reports. The effect of treatment with no prior reports is 0.75 (0.29,
1.88) p=0.539; the effect of treatment with one or more prior reports is 4.39 (0.50, 38.37)
p=0.181. Substantiation is again such a rare event that estimates are hugely imprecise.
Conclusion
Based on these analyses, the ARS-South Hayward program did not appear to have a
statistically significant effect on outcomes related to child welfare system involvement.
When holding the effect of prior reports constant, there is trend toward positive effect of
services on the outcomes of re-report for families with prior child welfare system
involvement. It has been suggested that the initial child maltreatment report may have an
inhibiting effect on subsequent re-report for a period of time (English et. al., 1999). The
small number of ARS clients without prior reports (n=16, 10%) makes any conclusion
based on this portion of the sample questionable. For the portion of the treatment sample
with prior reports, the trend toward a positive effect of treatment on re-reports is an
encouraging sign. However, treatment had no significant effects on investigation or
substantiation of re-reports.
142
Findings of a weak effect of treatment on subsequent child maltreatment are in line
with the literature on child maltreatment prevention and differential response
interventions. In Geeraert and associates' (2004) meta-analysis of 19 studies, and in
MacLeod and Nelson's (2000) meta-analysis of over 50 programs, positive results were
found, but the overall effect size (about 0.20 in both reviews) would be considered small
by conventional standards. A majority of studies conducted on differential response have
found that re-report rates are similar six months after treatment for families who receive
treatment and comparison groups (Center for Child and Family Policy, 2004; English et
al., 2000; U.S. Department of Health and Human Services, 2005).
The ARS intervention may have achieved beneficial outcomes with regards to the
proximal goals, such as making families feel more connected to their communities.
Certainly, the process portion of the study (Chapter 5) made clear that families appreciate
ARS services and perceive a number of benefits associated with the treatment. However,
the study design did not allow for examination of the program's proximal outcomes.
143
CHAPTER 8: DISCUSSION
Child welfare agencies across the state of California and throughout the nation are
undergoing a historic reform process intended to promote safety and permanency of
children through early intervention. While united under the umbrella term "differential
response," program models vary dramatically along several dimensions. One is the
population served, which may be families screened out of investigation, families referred
at the hotline to a "family assessment track," or families whose allegations of
maltreatment are found to be unsubstantiated. Another is who provides the service—a
community worker or CPS staff member—and whether they have professional
credentials or have paraprofessional status. Programs also differ by type of services
offered (e.g., assistance with basic needs, attention to the parent child relationship) and
whether services are offered in the home. The intensity and duration of services also vary
by program. Child welfare seems to be following Mao Tse-Tung's directive: "Let a
thousand flowers bloom." There may be advantages in allowing local jurisdictions to
create the differential response that best suits their needs. Yet this diversity of program
models can lead to a lack of coherence and difficulties in building an evidentiary base.
This dissertation research examined the Another Road to Safety program, which
can be characterized as a differential response model serving families screened out of
investigation, with four main types of interventions provided in-home by
paraprofessional community workers. Three strands of inquiry made up this study: a
qualitative portion that focused on the ARS program model and experiences of staff and
clients with service delivery; a Geographic Information Systems (GIS) portion that
addressed availability of services by neighborhood; and an outcomes portion that
examined child welfare systems involvement post-intervention for clients in contrast to a
comparison group. The purpose of this final chapter is to summarize findings across
methodologies and draw conclusions about potential implications. This is accomplished
in several parts. First, limitations to the research are described. Second, major findings
are reviewed, with reference to the current research literature. In conclusion,
implications are discussed for social welfare policy, social work practice, and future
research.
Limitations
Limitations apply to the research design and data collection, suggesting caution
when interpreting findings and drawing conclusions. The main concerns that emerged
have to do with sample construction, sample size, and the generalizability or
transferability of findings. Issues related to potential error, bias, and interpretation and
application of research findings are discussed, with methodologies for each research
question treated separately.
Data for question one, regarding staff and client experiences with the ARS
program, may have been compromised by the sampling procedure and biases in
participants' responses. Recruitment was handled by ARS staff, who may have neglected
to inform difficult clients of the study or more strongly encouraged participation by
enthusiastic clients. Client participation was self-selected and may consequently be
biased in ways that are unknown. Those who chose to participate could have been
particularly positive or negative regarding their experiences. Clients' responses may
have been subject to social desirability bias whereby they told the researcher what they
believed was expected, or were less than honest out of concern for their ARS worker's
145
position. While current clients were selected to minimize recall bias, there were
questions that asked clients to think back to their earliest experiences with the program.
Recollection of early experiences and opinions may have been influenced by later
program participation. Responses from workers may also have been influenced by social
desirability and recall biases. Although staff were assured that their comments were
confidential, they were also aware that administrators at the five partner agencies would
be provided with aggregated findings. The effect, if any, of this awareness on their
comments is unknown. While the qualitative aspect of the study offers valuable insight,
findings may not apply to other populations receiving other models of differential
response services. The reader can best judge the transferability of findings to other
contexts, based on an understanding of the research processes and assumptions applied in
this study.
The main limitations of the GIS portion of the study (Question 2) are related to
data integrity and point-in-time data. Data were gathered on nonprofit and public social
service locations and types from all sources known to the researcher. While due effort
was^made to ensure comprehensiveness, it is difficult to know how many services
relevant to families were actually captured by the analysis because a comprehensive
source of services in Alameda county does not exist. Moreover, data collection was
conducted at a single point in time, so it does not reflect neighborhood change. The same
study, replicated at a different time period, could result in different findings. Correlations
between child maltreatment rates and service density run on this data are cross-sectional
in nature, and causality cannot be inferred. It is possible that the observed correlation
146
between the two variables results from their connection to a third, omitted variable, rather
than a direct relationship.
The ability to draw strong conclusions from the outcomes portion of the study
(Question 3) is hampered by small sample sizes, which reduce the power of statistical
modeling for the data. Child maltreatment reports have a low base rate, even among high
risk populations (Guterman, 1997). Particularly when examining substantiation, a rare
outcome, small sample sizes make it difficult to detect a significant effect of treatment.
The findings related to substantiation in particular are indecisive at best.
Lack of control over assignment to treatment and the nature of the treatment
received also throw conclusions on the program's outcomes into question. While staff at
the Alameda County Department of Evaluation and Technology stated that comparison
group families were not offered the intervention simply due to program capacity, the
make-up of the two groups suggests selection bias. Almost all (90%) of the clients
referred to ARS-South Hayward who were included in the treatment sample had a history
of child welfare system involvement, whereas this was the case for only 66% of the
comparison group. It is possible that hotline screeners referred families to the ARS
program more frequently when they observed prior reports. Participation in ARS is
voluntary, so the sample may have been biased by self-selection. It is possible that
clients who were more troubled were more likely to opt for treatment, or alternatively,
that clients who were better prepared to change their parenting chose to participate. Non-
random assignment could have biased the sample in other unknown ways as well.
Intervention types and dosage were not controlled, as they might have been for a rigorous
randomized trial. While the lack of control over assignment and exposure to treatment
constitute threats to internal validity, these same conditions make the study more relevant
to the real-world, messy contexts in which differential response programs operate.
Another caveat has to do with the outcomes selected. Child welfare system
involvement may not equate with child maltreatment prevention, raising the question of
construct validity. Child maltreatment has been described as an "iceberg phenomenon":
only a small portion of actual cases are visible to the system, while the majority remains
hidden. The proportion of "visible" cases may have been different between the treatment
and comparison group for reasons not directly related to the treatment. Increased
surveillance may occur in programs emphasizing weekly contact with a mandated
reporter and referral to community services staffed by mandated reporters (Guterman,
1997). Such programs may prevent maltreatment recurrence in some cases while
promoting early detection in others, but the early detection effects could mask the overall
beneficial impact of services (Olds & Kitzman, 1993). At the same time, surveillance
bias is not a catch-all excuse for null findings. Studies which have accounted for this
source of bias in statistical modeling have typically found that its unique contribution is
small (Chaffin, 2005).
The administrative data used to examine client outcomes is prone to certain types
of errors and limitations that affect the research. As was mentioned in chapter 7, some
hand coding was done to identify ARS recipients in the CWS/CMS database. When
client identifiers were not available, child names and demographic descriptors were used
to make the link. Cases could have been misidentified during this process. Further, 42
clients identified by the ARS ECChange database as having been retained for services
could not be located in the CWS/CMS database and had to be dropped from the analysis.
Perhaps even more importantly, reliance on administrative child welfare data limits the
scope of analysis. Only outcomes associated with child welfare system involvement
could be examined due to lack of data on the comparison group in 'softer' domains, such
as child health and parent-child interaction, that are tracked for ARS clients in
ECChange. Reduction of researcher control over data completeness and variable
measurement is characteristic of research on administrative child welfare datasets (Drake
& Jonson-Reid, 1999).
Keeping in mind the limitations discussed above, this study has identified a
number of key findings with significant implications for practice and policy related to
differential response. Discussion will now turn to these major findings.
Major findings
Based on interviews with staff, the ARS program appeared robust and faithful to
the original model. Clients expressed satisfaction with the program and described
changes in their material well-being and parenting abilities that hinted at program
success. However, participation in the ARS-South Hayward program did not affect
families' subsequent involvement with the child welfare system. Of the three, this
program site is considered the most mature and closest to the original model designed by
First 5 and the Social Services Agency of Alameda County. For this reason, it may be
extrapolated that client outcomes for the other program locations are similar (or perhaps
even worse). It is possible that program effectiveness is affected by neighborhood, as
there is evidence supporting variation in social service availability across the three
neighborhoods. A more comprehensive study of all three program sites, with data on
service referral and usage, might be able to capture differences by neighborhood. It is
also possible that the intervention achieves its proximal goals of increased connections
with community resources, provision of temporary social support, elimination of unmet
basic need, and improvement of parent-child relationship. These outcomes are not
measured directly by the study and may indeed improve, based on comments by clients.
What might explain the lack of outcome findings for the ARS program? There is
a clearly thought-out conceptual model (Figure 2). If service inputs do not lead to the
expected client outcomes, is this due to faulty assumptions underlying the logic model?
Research on child maltreatment prevention programs has been criticized for focusing
more on hypothesized mechanisms underlying child maltreatment than on the behaviors
associated with child maltreatment themselves (Chaffin, 2005). The same criticism could
be made of interventions. Rather than directly intervening in maltreating behaviors of
parents, the ARS program provides four interventions aimed at reducing social isolation,
lack of connection to community resources, unmet basic needs, and poor parenting. By
making improvements in these areas, services are expected to ultimately reduce child
maltreatment. However, the empirical evidence supporting associations between the
interventions provided and child maltreatment prevention are weak or lacking.
ARS staff and administrators described interventions related to basic needs as a
necessary precursor to deeper work with families. Parents struggling with survival may
not have the energy and focus to work on the relationship with their child. What is the
evidence that meeting families' basic needs will lead to a reduction in child
maltreatment? The effects of monetary or material assistance for families at-risk of child
welfare involvement have been examined by a handful of studies. A review of family
preservation and family support programs found that programs designed to help families
150
meet basic concrete needs were more effective at preventing recurrence of maltreatment
than programs which offered parenting and child development-oriented services (Chaffin
et al., 2001). The opposite finding was identified by a meta-analysis of child
maltreatment prevention home visiting programs; programs identified as having a
concrete needs component had a smaller effect size on improving family functioning than
home visiting programs without such services. One explanation for this finding may be
that offers of concrete aid are more frequently made in programs that serve families at
greater levels of crisis and poverty, and creating lasting change may be more difficult
with this population than with those families who have their basic needs satisfied
(MacLeod & Nelson, 2000). Child welfare clients who participated in the Illinois
"Norman" program, which provided direct cash payment and housing expenditures, were
found to have a reduced rate of out-of-home placement, fewer days in substitute care, and
a greater rate of reunification than families not receiving assistance (Eamon & Kopels,
2004). Participation in a home-based intervention focused on basic needs for families
who met criteria for risk of child neglect was found to significantly reduce risk factors,
notably in the areas of parenting and everyday stress, and improve protective factors,
such as parental competence, from baseline to case closure and from baseline to the six-
month follow-up. Families were assisted using a combination of referrals to community
providers and direct monetary assistance from an emergency fund, and were offered
assistance with emergency needs within one working day of the initial assessment
(DePanfilis & Dubowitz, 2005).
The findings on concrete aid have generally been favorable. Cash and material
assistance may make a difference in those cases where the help offered truly fits the
151
families' needs. Indeed, help acquiring needed equipment such as a crib, or assistance in
paying a bill, may be a more effective child maltreatment intervention than education on
parenting skills or child development (Chaffin et al , 2001). However, in cases of great
financial stress, a small handout or purchase of equipment may not tangibly improve the
plight of families. It is hard to say whether families will reap long-term benefits from the
ARS basic needs interventions. Assistance from the basic needs fund may defuse a crisis
that occurs during program participation, but what of later crises? Similar basic needs
funds are not available from community providers. Families may be able to meet their
basic needs more sustainably from the knowledge they gain of community providers and
the skills they develop in accessing resources. However, it is unknown whether families
continue to access social services after program completion. This is questionable given
administrators' assumptions that families need nine months of intensive case
management to get the services they require.
Poor attachment between parents and children is hypothesized by the ARS
program to be a contributor toward the issue of child maltreatment. Is this assumption
supported by the literature, and what types of interventions have demonstrated success in
strengthening attachment? There is evidence that maltreated children are less securely
attached to their parents than non-maltreated children (Morton & Browne, 1998).
Moreover, attachment appears to be malleable, particularly early in life. Changing the
stability of the mother and child's life can translate into improvements in the attachment
relationship (Egeland & Sroufe, 1981). A meta-analysis of 16 intervention studies that
aimed to improve parental sensitivity and parent-child attachment found that short-term,
focused programs appear to have greater efficacy in affecting parental sensitivity and
152
parent-child relationship than longer term, broad-based program. Behaviorally-oriented
approaches that provide modeling and/or feedback on video-taped parent-child
interactions achieve greater success, at least in the short-term, than therapeutic
approaches. Interventions with a primary focus on urgent survival needs appear to have a
hard time simultaneously promoting maternal sensitivity, perhaps because such needs
trump attention to caregiving (van Ijzendoorn, Juffer, & Duyvesteyn, 1995).
While interventions to promote attachment are promising, their connection to
preventing child maltreatment is limited. In the context of ARS, this program component
was perhaps the weakest link in the program's service array. On the whole, ARS workers
seemed hesitant to directly address parenting. The approach to working with parents
around their parenting skills is not well-defined or consistent, lacking clear guidelines or
goals. Clients described receiving assistance with discipline and child behaviors, not
feedback intended to improve parental sensitivity. ARS administrators might consider re
evaluating this component of the program, and potentially adding targeted behavioral
interventions to provide feedback and modeling related to sensitive caregiving. While
they may not remediate problems in the attachment relationship, providing families with
fun opportunities like the Lawrence Hall of Science program may indirectly support the
development of positive relationships. Clients expressed their appreciation and
enjoyment of this aspect of the program.
ARS seeks to reduce social isolation and emotional stress through the mechanism
of perceived support. The regular contact and the warm relationship established between
families and their worker is intended to provide a general sense that help is available
when needed. Families describe feeling calmer and less stressed as a result of their
153
participation in the program. Is there reason to believe that families would consequently
be less likely to maltreat their children? A number of studies support the proposition that
maltreating parents measure lower on perceived support than parents who do not maltreat
their children (Daniel, Hampton, & Newberger, 1983; Egeland & Brunquell, 1979;
Kotelchuck, 1982; Newberger et al., 1986; Polansky et al., 1981; Polansky, Ammons, &
Gaudin, 1985; Turner & Avison, 1985). Yet the directionality and causality of this
relationship, and its implications, are unclear.
There are several possible scenarios to explain the connection between child
maltreatment and social isolation. Families may lack in social supports, in which case it
may be logical to increase their social networks. There is some evidence that maltreating
families identify fewer social contacts that are reliable or potential sources of support
(Kotelchuck, 1982; Polansky et al., 1981; Polansky, Ammons, & Gaudin, 1985).
Another possible explanation is that families have sufficient networks but fail to make
use of the supports available to them, in which case social skills training may make more
sense. A few studies have identified behavioral patterns that suggest distrust of society
and parental preference for sequestering the family and handling problems independently
(Elmer & Gregg, 1967; Young, 1964). A third possibility is that perceived social support
is a stable personality trait (Lakey & Lutz, 1996) that is not malleable for intervention.
Garbarino (1977) points out that social isolation, like most human phenomenon, is
complex and multiply determined. As yet it is unclear which social support functions
maltreating parents lack, and therefore which types of interventions could be most helpful
(DePanfilis, 1996). However, there does not appear to be any empirical support for the
assumption that grafting a temporary relationship with a worker onto families' lives will
154
ameliorate a low sense of perceived support in the long-term. A randomized control
study of high-risk mothers provided with home visitation or no service found significant
differences in overall social support, particularly in the domains of affective support,
affirmation, and availability of concrete assistance; however, the researchers speculated
that these gains came from the relationship with the home visitor alone, not from changes
in the mother's social network (Marcenko & Spence, 1994). Overall social support for
mothers participating in Hawaii's Healthy Start perinatal home visitation program did not
increase as a result of participation in the program, as compared to a control group.
There was a difference between the groups in perceived emotional support from a close
adult; since the social network of participants did not appear to change, this support likely
came from the home visitor (McCurdy, 2001). If increased social support is the goal, a
more sustainable intervention than the relationship with the home visitor must be
considered.
Referral to community resources is, perhaps, the greatest focus of attention by the
ARS program. Why would administrators expect that this service might reduce the
likelihood of future child maltreatment? The presumed mechanism underlying the
connection between institutional resources and child maltreatment is unclear. There does
appear to be some relationship between neighborhood conditions and parents'
perceptions and usage of community resources (Coulton, 1996). Parents' perceptions of
community resources may further be connected to parenting practices related to
proactively finding positive opportunities for children (Elder, et al., 1995) as well as
discipline and harsh behavior (Pinderhughes, Nix, Foster & Jones, 2001). However, ARS
does not engage in activities to transform the array of neighborhood social services;
155
rather, the program shares information on resources with families and helps facilitate
their connection to services. Connection to community resources seems intended to meet
two of the goals previously described: provision of support through formal social
services, and access to resources to meet basic needs.
As such, this type of ARS intervention is subject to similar criticism as the social
support and basic needs strategies. The focus is exclusively on formal sources of social
support and connections to formal institutions in communities, rather than informal
connections to support and resources from peers and neighborhood groups. Exclusively
stressing formal providers and resources could potentially attenuate informal ties in
communities (McKnight, 1995), a source of support that is generally preferred over
formal sources.
With regard to the theories underpinning the four main programmatic
interventions, support on the connection between child maltreatment and connection to
institutional resources is weakest. Attachment and child maltreatment appear to be
connected, but it seems that more sensitive parenting leads to better attachment, not that
better attachment leads to improved parenting (and thus less maltreatment). A growing
body of evidence supports the provision of basic needs assistance, but the duration and
amount of assistance needed to make a difference is not known. The evidence is
strongest for an association between social support and child maltreatment, but even here,
the literature does not support temporary relationships with workers. The meaning of
these findings will next be explored, for social welfare policy, social work practice, and
future research.
156
Implications for social welfare policy
There is neither strong empirical evidence nor robust theoretical support
connecting the ARS program's four main interventions to the stated goal of child
maltreatment prevention. Unfortunately, weak or neutral findings for programs seeking
to prevent the occurrence or reoccurrence of child maltreatment are the rule rather than
the exception. In Geeraert and associates' (2006) meta-analysis of 19 child maltreatment
prevention studies, and in MacLeod and Nelson's (2000) meta-analysis of over 50,
positive results were found, but the overall effect size on reducing maltreatment averaged
only about 0.20 in both reviews — an effect size considered small by conventional
standards (Cohen, 1988). A majority of studies conducted on differential response have
found that re-report rates are similar six months after treatment for families who receive
treatment and comparison groups (Center for Child and Family Policy, 2004; English et
al., 2000; U.S. Department of Health and Human Services, 2005). The field of child
welfare may need to be more realistic about what can be expected from a limited
intervention regarding child maltreatment outcomes. Prevention of child abuse may
require multi-modal, intensive services. The field is still searching for effective means of
preventing child maltreatment occurrence and recurrence.
This study brings up the question of whether the child welfare system should be
focusing its resources and attention on evaluated out or unsubstantiated cases. Looking at
the population of California as a whole, re-referral among initially evaluated out cases is
a low base rate phenomenon (Needell et al., 2005), making it difficult to demonstrate
intervention effects and casting doubt on whether addressing this issue is the best use of
scarce resources. However, as this and other studies (Drake et al., 2003; Wolock, 2001)
157
demonstrate, when you give families enough time, re-report is a problem—in this study,
affecting 30% of cases. This volume of re-referral suggests that families screened out of
investigation are troubled, and may need services and connection to ongoing support.
Programs such as ARS program may do better on outcomes related to child well-
being and family support than child maltreatment prevention. Family support programs
usually do offer modest benefits to families related to parenting attitudes, knowledge, and
behavior as well as family functioning (Powell, 1994). Based on the model and client
comments, it can be surmised that something beneficial is happening for families. ARS
administrators will have to decide if these types of outcomes justify investment in the
program. Indeed, this is also a question for the broader field of child welfare.
Differential response, with its emphasis on voluntary, strengths-based, community
support for families suggests a family support approach (Tilbury, 2005). To justify
public expenditure, particularly the diversion of funds from other child welfare activities
through the Title IV-E waiver, child welfare administrators must clarify the purpose of
differential response. If it is family support, then the field will have to be content with
the goal of strengthening families who may have never entered the system as well as the
portion who do (between 7-50%, depending on the study and the timeframe used). If the
goal is child maltreatment prevention and fewer families entering the formal child
welfare system, then services will need to be targeted and program models scrutinized to
determine whether there is a clear causal chain between services delivered and outcomes
expected. The enthusiasm for this approach could easily turn into cynicism if promises
are not borne out by research findings, as was the case for an earlier child welfare "silver
bullet"—family preservation services.
158
Differential response is part of a host of initiatives intended to reform the child
welfare system by involving communities in child protection. Descriptions of DR.
emphasize community partnerships and reliance on community-based services, yet the
neighborhood context tends to receive little attention in program planning and policy.
This omission is surprising, given that the primary intervention is to connect families to
community services. GIS findings from this study suggest that regional variation in
resource distribution is a factor that may influence program implementation. Policy
makers and program administrators might consider a number of steps to better account
for the neighborhood context. Prior to implementing DR, community assessments may
be useful for identifying what services are available, gaps, and accessibility challenges.
In particular, child welfare agencies would benefit from looking at the match between
client needs and available services. Using GIS to map services can help supply this
information. To have a truly sustainable impact on families' lives, child welfare agencies
might consider taking a leadership role to improve local service arrays, thereby
improving availability and quality of services, and potential client outcomes.
Comprehensive assessment of clients' experiences with DR would naturally include
assessment of their experiences with referrals—whether they followed up on referrals,
barriers they may have experienced, and their experiences and satisfaction with services
from referrals. Without this information, there is an incomplete picture of the effects of
DR programs.
Implications for social work practice
ARS may not yet achieve its ultimate goal of child maltreatment prevention. Yet
there are common sense reasons why the programmatic interventions might benefit
159
families, even if the connection with positive outcomes is not documented in the research
literature. Interventions in the areas of basic needs, the parent-child relationship, social
support, and connection to institutional resources may contribute to strengthening
families and helping them weather crises. A reconsideration of the ARS program model
and the logical connections to proximal and distal goals is recommended. Ideas for
improving the four main ARS interventions are offered below.
ARS seeks to address the three lowest levels of the Maslow Hierarchy: basic
physiological needs, safety, and love. These needs could be addressed in a more
concerted fashion if Maslow theory was brought more to the fore and families were
specifically assessed on these types of needs. Examination of these need categories are
currently folded into other assessment tools, but one idea would be to create a tool which
breaks out needs according to Maslow's hierarchy. Assessment of physiological needs
could encompass housing, food, clothing, and employment. Attention to safety needs
would involve that of family members within the home—for example, household hazards
for children and domestic violence—as well as safety within the community—such as
exposure to theft and gang violence. Evaluation of families' need for love might involve
assessment of family dynamics and networks of extended family and friends. Needs
could be addressed sequentially, as per the Maslow hypothesis that satisfaction of lower
needs can prepare people to address higher needs. This hypothesis could be tested
informally, through staff observations regarding clients' progress, or more formally using
psychosocial assessments (though the research reviewed in chapter 3 suggests that
current tools imperfectly capture the concepts underlying the Maslow Hierarchy).
The parent-child relationship is a critical, yet sensitive, area for intervention.
Trust must first be established between worker and family. Once families are receptive
to feedback, research suggests that behavioral interventions focused on promotion of
caregiver sensitivity show the most promise in improving the attachment relationship. To
implement such an approach, the ARS paraprofessional staff would need training on how
to offer feedback to parents and how to model appropriate parenting behaviors. Perhaps
such training would overcome their hesitancy in discussing parenting issues. It is also
possible that a more highly educated staff would achieve greater success in addressing
the parenting domain. Administrators might reconsider the choice to employ
paraprofessionals, and whether these staff are equipped to help families improving their
parental sensitivity and skills.
Being available and encouraging, staff may bolster clients' perceived support,
which in turn can buffer the effects of stress. Clients clearly appreciated the support they
received from the worker; indeed, many appeared to rely upon it. This raises a concern
about the sustainability of social support. A program component focused on building
families' social networks and social skills might be considered, to make last changes in
the availability of social support for families. A natural place to start would be to
encourage the development of relationships between families enrolled in ARS. Indeed,
staff suggest this has often occurred as a by-product of the Lawrence Hall of Science
Program. At least one of the ARS sites was also considering the development of a
mothers' support group. In addition, ARS workers could help families work on their
existing relationships, by encouraging them to reach out to extended family, friends, and
neighbors. The field of child welfare has embraced the notion that extended family and
161
friends should be involved in making decisions for children involved in the system. If
this is feasible for families whose issues require court-mandated involvement with child
protection, surely the same can be done for families who are participating in a voluntary
program.
Teaching families about available resources in the communities and how to access
them can be a way to implant knowledge and skills that allow families to meet future
needs independently. Yet as family conditions change and children grow, new issues
may arise in families, necessitating different types of services and support. Gains that
families make while in the program may require an occasional booster in order to be
maintained, as new crises surface. The ARS program might consider designating a staff
person to maintain contact with families and continue to assist them beyond the
program's duration.
Future research
From this research, certain broader questions have emerged about the differential
response approach. First and foremost, what types of program models may be most
effective in reaching the goal of preventing low to moderate risk families from entering
the child welfare system? Different jurisdictions have taken a number of approaches, and
it is time to identify and study successful models. Second, what are successful strategies
for engaging clients in DR services? Client engagement is a challenge because the point
of entry for DR is a child abuse report, and many families do not accept services out of
suspicion of child protective services. This is another case where promising practices can
be identified and shared with the field. Third, how should the outcomes of DR services
be assessed? As was discussed in the limitations section, re-report of child maltreatment
162
may be a flawed measure if families served by DR continue to make greater use of
community services after their completion of DR services, putting them under increased
surveillance. The field of child welfare needs to clarify the goals of DR and find
appropriate means to measure them. Beyond simply reducing maltreatment, DR seeks to
build protective factors, and these would be important to measure. Because
administrative child welfare data are limited, new methodological designs and methods of
data collection must be devised.
An evidentiary base on the intensity, duration, targeting, staffing, and content of
differential responses services is needed to inform policy and practice. Much of the
current literature in this area is descriptive, not empirical. The Administration for
Children, Youth, and Families (Children's Bureau) has recently announced funding for
the development of the "National Quality Improvement Center on Differential Response
in Child Protective Services." The purpose of this center will be to generate knowledge
on differential response practice models and to support infrastructure development at
state and local levels for implementation of services. This move toward building the
knowledge base around DR is exactly what is needed, as the field of child welfare begins
to separate the hopes from the realities in terms of what can be achieved through offering
preventative, voluntary, strengths-based community services to families screened out of
traditional child welfare services.
163
References
Aber, J.L., Gephart, M.A., Brooks-Gunn, J. and Connell, J.P. (1997). Development in
context: Implications for studying neighborhood effects. In J. Brooks-Gunn, G.J.
Duncan, & J.L. Aber (Eds.), Neighborhood poverty, Vol. 1: Context and
consequences for children (pp. 79-118). New York: Russell Sage Foundation.
Ainsworth, M. (1973). The development of infant-mother attachment. In B. Caldwell &
H. Ricciuti (Eds.), Review of child development research. Vol. 3. Chicago:
University of Chicago Press.
Ainsworth, M. D. S., Blehar, M. C, Waters, E., & Wall, S. (1978) Patterns of
attachment: a psychological study of the strange situation. Hillsdale, N.J.:
Erlbaum.
Allison, P.D. (1984). Event history analysis. Quantitative Applications in the Social
Sciences Series. M.S. Lewis-Beck (Ed.). Newbury Park: Sage Publications.
Baird, C. & Wagner, D. (2000). The relative validity of acturial- and consensus-based
risk assessment systems. Children and Youth Services Review, 22 (11/12), 839-
871.
Baird, C, Wagner, D., Healy, T., & Johnson, D. (1999). Risk assessment in child
protective services: Consensus and actuarial model reliability. Child Welfare, 78
(6), 723-748.
Barker, C, & Pistrang, N. (2002). Psychotherapy and social support: Integrating research
on psychological helping. Clinical Psychology Review, 22, 361-379.
Barrera, M. (1986). Bronfenbrenner, U. (1979). The ecology of human development.
Cambridge, MA: Harvard University Press.
Barth, R. P. (1991). An experimental evaluation of in-home child abuse prevention
services. Child Abuse & Neglect, 15, 363-375.
Bilukha, O., Hahn, R.A., Crosby, A., Fullilove, M.T., Liberman, A., Oscicki, E.,
Snynder, S., Tuma, F., Corso, P., Schofield, A., & Briss, P.A. (2005). The
effectiveness of early childhood home visitation in preventing violence: A
systematic review. American Journal of Preventive Medicine, 28 (2S1), 11-39.
Bolen, R. (2000). Validity of attachment theory. Trauma, Violence, & Abuse, 1 (2), 128-
153.
Boss, P. (1987). Family stress: Perception and context. In M. Sussman & S. Steinmetz
(Eds.), Handbook on marriage and the family (pp. 695-723). New York: Plenum.
Bowlby, J. (1969), Attachment and loss, Vol. 1: Attachment. New York: Basic Books.
Bowlby, J. (1988). A secure base: Parent-child attachment and healthy human
development. New York: Basic Books.
Bronfenbrenner, U. (1974). Is early intervention effective? Early Childhood Education
Journal, 2 (2), 14-18.
Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature
and design. Cambridge, Massachusetts: Harvard University Press.
Brooks-Gunn, J. McCormick, M.C., Klebanov, P,K. & McCarton, C. (1998). Health care
use of 3-year-old low birth weight premature children: Effects of family and
neighborhood poverty. Journal of Pediatrics, 132, 971-5.
Burton, L.M. & Jarrett, R.L. (November 2000). In the Mix, Yet on the Margins: The
Place of Families in Urban Neighborhood and Child Development Research.
Journal of Marriage and the Family, 62, 1114-1135.
165
Carlson, V., Cicchetti, D., Baraett, D. & Braunwald, K. (1989). Disorganized/disoriented
attachment relationships in maltreated infants. Developmental Psychology 25,
525-531.
Center for Child and Family Policy (2004). Multiple Response System (MRS) evaluation
report to the North Carolina Division of Social Services (NCDSS). Retrieved
online March 15, 2005 from the Center for Child and Family Policy web site:
http://www.pubpol.duke.edu/centers/child
Chaffin, M. (2005). Response to letters. Child Abuse & Neglect, 29, 241-249.
Chaffin, M., Bonner, B. L., & Hill, R. F. (2001). Family preservation and family support
programs: Child maltreatment outcomes across client risk levels and program
types. Child Abuse and Neglect, 25, 1269-1289.
Chase-Lansdale, P.L., Gordon, R.A., Brooks-Gunn, J. & Klebanov, P.K. (1997).
Neighborhood and family influences on the intellectual and behavioral
competence of preschool and early school-age children. In J. Brooks-Gunn, G.J.
Duncan, & J.L. Aber (Eds.), Neighborhood poverty, Vol. 1: Context and
consequences for children (pp. 79-118). New York: Russell Sage Foundation.
Chaskin, R.J., Brown, P., & Venkatesh, S. (2001) Building community capacity. New
York: Aldine de Gruyter.
Child Welfare Stakeholders Group (2003). CWS redesign: The future of California's
child welfare services final report. Sacramento. Retrieved online 12 March 2005
from the California child welfare services redesign web site:
http://www.cwsredesign.ca.gov/res/pdf/CWSReport.pdf
166
Christensen, A., & Jacobson, N. S. (1994). Who (or what) can do psychotherapy? The
status and challenge of nonprofessional therapies. Psychological Science, 5, 8-14.
Cohen, J. (1988, revised edition). Statistical power analysis for the behavioral sciences.
New York: Academic Press.
Coleman, J.S. (1968) Equality of educational opportunity. Equity & Excellence in
Education, 6(5), 19-28.
Collett, D. (1994). Modelling survival data in medical research. Texts in Statistical
Science Series. C. Chatfield and J. V. Zidek (Eds.). New York: Chapman & Hall.
Connell, J.P. & Halpern-Felsher, B.L. (1997). Neighborhood and family influences on the
intellectual and behavioral competence of preschool and early school-age
children. In J. Brooks-Gunn, G.J. Duncan, & J.L. Aber (Eds.), Neighborhood
poverty, Vol. 1: Context and consequences for children (pp. 79-118). New York:
Russell Sage Foundation.
Coulton, C. (1996). Effects of neighborhoods on families and children: Implications for
services. In A.J. Kahn & S.B. Kammerman (Eds.), Children and their families in
big cities: Strategies for service reform (pp. 87-120). New York: Cross-National
Studies Research Program.
Cowen, E.L. (1982). Help is where you find it: Four informal helping groups. American
Psychologist, 37(4), 385-395
Daniel, J.H., Hampton, R.R. and Newberger, E.H., 1983. Child abuse and accidents in
Black families: A controlled comparative study. American Journal of
Orthopsychiatry 53, 645-653.
167
Darling, N. & Steinberg, L. (1997). Community influences on adolescent achievement
and deviance. In J. Brooks-Gunn, G.J. Duncan, & J.L. Aber (Eds.), Neighborhood
poverty, Vol. 2: Policy implications in studying neighborhoods (pp. 23-47). New
York: Russell Sage Foundation.
Daro, D., McCurdy, K., & Nelson, C. (2005). Engagement and retention in voluntary
parent support programs: Final report. Retrieved online November 2, 2005 from
the Chapin Hall Center for Children web site:
http://www.chapinhall.org/article abstract.aspx?ar=1410&L2=60&L3=126
DePanfilis, D. (1996). Social isolation of neglectful families: A review of social support
assessment and intervention models. Child Maltreatment, 1 (1), 37-52.
DePanfilis, D. & Dubowitz, H. (2005). Family connections: A program for preventing
child neglect. Child Maltreatment, 10(2), 108-123.
Diez Roux, A.V. (2001, Nov.) Investigating neighborhood and area effects on health.
American Journal of Public Health, 91 (11), 1783-1789.
Drake, B. & Jonson-Reid, M. (1999). Some thoughts on the increasing use of
administrative data in child maltreatment research. Child Maltreatment, 4, 308-
315
Drake, B., Jonson-Reid, M., Way, I., Chung, S. (2003). Substantiation and recidivism.
Child Maltreatment, 8 (4), 248-260.
Duggan, A., McFarlane, E., Fuddy, L., Burrell, L., Higman, S.M., Windham, A., & Sia,
C. (2004). Randomized trial of a statewide home visiting program: Impact in
preventing child abuse and neglect. Child Abuse & Neglect, 28, 597-622.
168
Duggan, A. K., McFarlane, E.C., Windham, A.M., Rohde, C.A., Salkever, D.S., Fuddy,
L., Rosenberg, L.A., Buchbinder, S.B., & Sia, C.C.J. (1999). Evaluation of
Hawaii's Healthy Start Program. The Future of Children, 9(1), 66-90.
Dubow, E.F., Edwards, S., Ippolito, M.F. (1997). Life stressors, neighborhood
disadvantage, and resources: A focus on inner-city children's adjustment. Journal
of Clinical Child Psychology, 26 (2), 130-144
Durlak, J. (1979). Comparative effectiveness of paraprofessional & professional helpers.
Psychological Bulletin, 86, 80-92.
Eamon, M. K. & Koppel, S. (2004). 'For reasons of poverty': Court challenges to child
welfare practices and mandated programs. Children & Youth Services Review, 26,
821-836.
Eckenrode, J., & Hamilton, S. (2000). One-to-one support interventions: Home visiting
and mentoring. In S. Cohen, L. Underwood, & B. Gottlieb (Eds.), Social support
measurement and interventions: A guide to health and social scientists. New
York: Oxford University Press.
Egeland, B. and Brunquell, D. (1979). An at-risk approach to the study of child abuse:
Some preliminary findings. American Academy of Child Psychiatry 18, 219-235.
Egeland, B. & Sroufe, L.A. (1981). Attachment & early maltreatment. Child
Development, 52, 44-52.
Elder, G.H., Eccles, J.S., Ardelt, M. & Lord, S. (Aug., 1995). Inner-city parents under
economic pressure: Perspectives on the strategies of parenting. Journal of
Marriage and the Family, 57 (3), 771-784.
169
Ellen, I.G. & Turner, M.A. (1997). Does neighborhood matter? Assessing recent
evidence. Housing Policy Debate, 8 (4), 833-866.
Elliott, D.S., Wilson, W.J., Huizinga, D., Sampson, R.J., Elliott, A., & Rankin, B. (1996).
The effects of neighborhood disadvantage on adolescent development. Journal of
Research in Crime and Delinquency, 33, 389-426.
Elmer, E. and Gregg, G. (1967). Developmental characteristics of abused children.
Pediatrics 40, 98-102.
English, D. J., Marshall, D. B., Brummel, S., & Orme, M. (1999). Characteristics of
repeated referrals to child protective services in Washington state. Child
Maltreatment, 4(4), 297-307.
English, D. J., Wingard, T., Marshall, D., Orme, M., & Orme, A. (2000). Alternative
responses to child protective services: emerging issues and concerns. Child Abuse
and Neglect, 24 (3), 375-388.
Ensminger, M.E., Lamkin, R.P. & Jacobson, N. (1996). School leaving: A longitudinal
perspective including neighborhood effects. Child Development, 67, 2400-2416.
Environmental Sciences Research Institute (ESRI) (Oct 31, 2006). Coverage. GIS
Dictionary. Retrieved online April 17, 2008 from ESRI web site:
http://support.esri.com/index.cfm?fa=knowledgebase.gisDictionary.search&searc
h=true&searchTerrn=coverage
Finfgeld-Connett, D. (March 2005). Clarification of social support. Journal of Nursing
Scholarship, 37 (1), 4-9.
First 5 Alameda County Staff (2004). Administration for Children, Youth and Families,
2003-2004 Final Report for Another Road to Safety. Unpublished document.
170
Foundation Consortium (2005). Differential Response in California. Retrieved online 9
September 2005 from the Foundation Consortium website:
http://www.foundationconsortium.org/pdf/cws/dr_overview.pdf
Freisthler, B. (2004). A spatial analysis of social disorganization, alcohol outlets, and
rates of child maltreatment in neighborhoods. Children and Youth Services
Review, 26, 803-819.
Fuller, B., Coonerty, C, Kipnis, F. & Choong, Y. (1997, Nov.) ,4?? unfair head start:
California families face gaps in preschool and child care availability. Berkeley,
CA: Growing Up in Poverty Project.
Fuller, B., Kagan, S.L., Loeb, S. & Chang, Y. (2004) Child care quality: Centers and
home settings that serve poor families. Early Childhood Research Quarterly 19,
505-527.
Furstenberg, F.F. & Hughes, M.E. (1997). The influence of neighborhoods on children's
development: A theoretical perspective and a research agenda. In J. Brooks-
Gunn, G.J. Duncan, & J.L. Aber (Eds.), Neighborhood poverty, Vol. 2: Policy
implications in studying neighborhoods (pp. 23-47). New York: Russell Sage
Foundation.
Garbarino, J. (1977). The Human Ecology of Child Maltreatment: A Conceptual Model
for Research. Journal of Marriage and the Family, 39 (4) 721-735.
Garbarino, J. & Kostelny, K. (1993). Neighborhood and community influences on
parenting. In T. Luster & Okagaki, L. (Eds.), Parenting: An ecological
perspective. Hillsdale, NJ: Lawrence Erlbaum Associates.
Garbarino, J. & Sherman, D, (1980, Mar.) High-risk neighborhoods and high-risk
families: The human ecology of child maltreatment. Child Development, 51 (1),
188-198.
Garner, C.L., Raudenbush, S.W. (1991, Oct.) Neighborhood effects on educational
attainment: A multilevel analysis. Sociology of Education, 64 (4), 251-262.
Geeraert, L., Van den Noortgate, W., Grietens, H., & Onghena, P. (2004). The effects of
early prevention programs for families with young children at risk for physical
child abuse and neglect: A meta-analysis. Child Maltreatment, 9 (3), 277-291.
George, C, Kaplan, N., & Main, M. (1985). The Berkeley adult attachment interview.
Unpublished manuscript, Department of Psychology, University of California,
Berkeley.
Gephart, M.A. (1997) Neighborhoods and communities as contexts for development. In
J. Brooks-Gunn, G.J. Duncan, & J.L. Aber (Eds.), Neighborhood poverty, Vol. 1:
Context and consequences for children (pp. 1-43). New York: Russell Sage
Foundation.
Guterman, N. (1997). Early prevention of physical child abuse and neglect: Existing
evidence and future directions. Child Maltreatment, 2 (1), 12-34.
Guterman, N. (1999). Enrollment strategies in early home visitation to prevent physical
child abuse and neglect and the "universal versus targeted" debate: A meta
analysis of population based and screening-based programs. Child Abuse &
Neglect, 23 (9), 863-890.
Guterman, N. (2001). Stopping child maltreatment before it starts: Emerging horizons in
early home visitation services. Thousand Oaks: Sage Publications, Inc.
172
Hahn, R. A., Mercy, J., Bilukha, O. & Briss, P. (2005). Assessing home visiting programs
to prevent child abuse: Taking silver and bronze along with gold. Child Abuse &
Neglect, 29 (3), 215-218.
Hattie, J. A., Sharpley, C. F., & Rogers, H. J. (1984). Comparative effectiveness of
professional and paraprofessional helpers. Psychological Bulletin, 95, 534-541.
Hiatt, S. W., Sampson, D., & Baird, D. (1997). Paraprofessional home visitation:
Conceptual and pragmatic considerations. Journal of Community Psychology, 25,
77-93.
Hinson Langford, C.P., Bowsher, J., Maloney, J.P. & Lillis, P.P. (1997). Social support:
A conceptual analysis. Journal of Advanced Nursing, 25, 95-100.
Hogan, B.E., Linden, W. & Najarian, B. (2002). Social support interventions: Do they
work? Clinical Psychology Review 22, 381^440.
House, J. S. (1981). Work stress and social support. Reading MA: Addison-Wesley.
Howe, D., Brandon, M., Hinings, D. & Schofield, G. (1999). Attachment theory, child
maltreatment, and family support. Mahwah, New Jersey: Lawrence Erlbaum
Associates.
Hoyle, R.H., Harris, M.J. & Judd, CM. (2002) Research methods in social relations, 7th
edition. Belmont, CA: Wadworth/Thomas Learning Inc.
Ihlanfeldt, K.R. & Sjoquist, D.L. (1998). The spatial mismatch hypothesis: A review of
recent studies and their implications for welfare reform. Housing Policy Debate, 9
(4), 849-892.
Inkelas, M. & Halfon, Neal. (1997). Recidivism in child protective services. Children and
Youth Services Review, 19(3), 139-161.
173
Jarrett, R. L. (1997). African American family and parenting strategies in impoverished
neighborhoods. Qualitative Sociology, 20, 275-288.
Jarrett, R.L. (1999, Fall). Successful parenting in high-risk neighborhoods. The Future of
Children, 9 (2), 45-50.
Jencks, C. & Mayer, S.E. (1990). The social consequences of growing up in a poor
neighborhood In Lynn, L.E. & M.G.H. McGeary (Eds.) Inner-city poverty in the
United States (pp.111-186). Washington, D.C.: National Academy Press.
Jung, J. (1987). Toward a social psychology of social support. Basic and Applied Social
Psychology, 8 (1& 2), 57-83.
Katz, L.F., Kling, J.R., & Liebman, J.B. (2001, May). Moving to Opportunity in Boston:
Early Results of a Randomized Mobility Experiment. The Quarterly Journal of
Economics, 116(2), 607-654.
Klebanov, P.K., Brooks-Gunn, J., McCarton, C, & McCormick, M.C. (Oct., 1998). The
contribution of neighborhood and family income to developmental test scores
over the first three years of life. Child Development, 69 (5), 1420-1436.
Korfmacher, J. (1999). Home visiting: Promise & peril. The Signal, 7, 1-8.
Kornhauser, R.R. (1978). Social sources of delinquency: An appraisal of analytic models.
Chicago: University of Chicago Press.
Kotelchuck, M. (1982). Child abuse and neglect: Prediction and misclassification. In R.
Starr (Ed.). Child abuse prediction (pp. 67-104), Ballinger, Cambridge, MA.
Lakey, B. & Lutz, C.J. (1996). Social support and preventive and therapeutic
interventions. In G.R. Pierce, B.R. Sarason, & I.G. Sarason (Eds.), Handbook of
social support and the family (pp. 435-464). New York: Plenum.
174
Lakey, B., Ross, L., Butler, C, & Bentley, K. (1996). Making social support judgments:
The role of perceived similarity and conscientiousness. Journal of Social and
Clinical Psychology, 15, 283-304.
Lawerence-Karski, R. (1999). Key decisions in child protective services: Report
investigation and court referral. Children and Youth Services Review, 21 (8), 643-
656.
Leventhal, T. & Brooks-Gunn, J. (2000). The neighborhoods they live in: The effects of
neighborhood residence on child and adolescent outcomes. Psychological
Bulletin, 126 (2), 309-337.
Loman, L.A. & Siegel, G.L. (2004a). Minnesota alternative response evaluation final
report. Retrieved online March 20, 2005 from the Institute of Applied Research
web site: http://iarstl.org/papers/ARFinalEvaluationReport.pdf
Loman, L. A. & Siegel, G.L. (2004b). Differential response in Missouri after five years.
Retrieved online April 5, 2005 from the Institute of Applied Research web site:
http://iarstl.org/papers/MODiffResp2004a.pdf
Lyons-Ruth, K., Repacholi, B., McLeod, S., & Silva, E. (1991). Disorganized attachment
behavior in infancy: Short-term stability, maternal and infant correlates, and risk-
related subtypes. Development andPsychopathology, 3, 377-396.
MacLeod, J., & Nelson, G. (2000). Programs for the promotion of family wellness and
the prevention of child maltreatment: a meta-analytic review. Child Abuse and
Neglect, 24 (9), 1127-1149.
Main, M. & Hesse, E. (1990). Parents' unresolved traumatic experiences are related to
infant disorganized attachment status: Is frightened and/or frightening parental
175
behavior the linking mechanism? In Greenberg M.T., Cicchetti D. & Cummings
E.M. (Eds.), Attachment in the preschool years (pp. 161-182). Chicago: Chicago
University Press.
Main, M. and Solomon, J. (1986). Discovery of an insecure disorganized/disoriented
attachment pattern: procedures, findings and implications for the classification of
behavior. In T. Braxelton and M.Yogman (Eds) Affective development in infancy,
(pp. 95-124). Norwood, NJ: Ablex
Main, M., & Weston, D. (1981). Security of attachment to mother and father: Related to
conflict behavior and readiness to establish new relationships. Child
Development, 52, 932-940.
Marcenko, M.O. & Spence, M. (1994). Home visitation services for at-risk pregnant and
postpartum women: A randomized trial American Journal of Orthopsychiatry 64,
468-478.
Marshall, D.B. & English, D.J. (1999). Survival analysis of risk factors for recidivism in
child abuse and neglect. Child Maltreatment, 4, 287-296
Maslow, A.H. (1943). A theory of human motivation. Psychological Review, 1943 (50),
370-396.
Maslow, A. H. (1970). Motivation and personality. New York: Harper and Row, 2nd ed.
McCroskey, J. & Meezan, W. (1998). Family-centered services: Approaches and
effectiveness. The Future of Children, 8 (1), 54-71.
McCurdy, K. (2001). Can home visitation enhance maternal social support? American
Journal of Community Psychology, 29(1): 97-112.
176
McCurdy, K. (2000). What works in nonmedical home visiting: Healthy Families
America. In M.P. Kluger, G. Alexander & P.A. Curtis (Eds.), What Works in
Child Welfare (pp. 45-56). Washington, D.C.: CWLA Press.
McGuigan, W. M., Katzev, A.R., & Pratt, C.C. (2003). Multi-level determinants of
retention in a home-visiting child abuse prevention program. Child Abuse &
Neglect, 27, 363-380.
McKnight, J. (1995). The careless society: Community and its counterfeits. New York:
Basic Books.
Morton, N. & Browne, K.D. (1998). Theory and observation of attachment and its
relation to child maltreatment: A review. Child Abuse & Neglect, 22 (11), 1093-
1104.
Musick, J. S. & Stort, F.M. (1990). Paraprofessionals, parenting, and child development:
Understanding the problems and seeking solutions. In S. Meisels & J. Shonkoff
(Eds.), Handbook of early intervention (pp. 651-667). Cambridge, U.K.:
Cambridge University Press.
Needell, B., Webster, D., Armijo, M., Lee, S., Cuccaro-Alamin, S., Shaw, T., Dawson,
W., Piccus, W., Magruder, J., Exel, M., Conley, A., Smith, J. , Dunn, A., Frerer,
K., Putnam Hornstein, E., & Kaczorowski, M.R. (2005). Child welfare services
reports for California. Retrieved November 2, 2005 from University of California
at Berkeley Center for Social Services Research web site:
http://cssr.berkeley.edu/CWSCMSreports/referrals/recurrence.asp
177
Newberger, E.H., Hampton, R.L., Marx, T.J. & White, K.M. (1986). Child abuse and
pediatric social illness: An epidemiological analysis and ecological reformulation.
American Journal of Orthopsychiatry 56, 589-601.
Olds, D. L. & Kitzman, H. (1993). Review of research on home visiting for pregnant
women and parents of young children. The Future of Children, 3 (3), 53-92.
Olds, D. L., Robinson, J., O'Brien, R., Luckey, D.W., Pettitt, L.M., Henderson, C.R., Ng,
R.K., Sheff, K.L., Korfmacher, J., Hiatt, S., & Talmi, A. (2002). Home visiting by
paraprofessionals and by nurses: A randomized, controlled trial. Pediatrics, 110
(3), 486-496.
Padgett, D.K. (1998). Qualitative methods in social work research: Challenges and
rewards. Thousand Oaks: Sage Publications.
Pebley, A.R. & Sastry, N. (2003, Feb.) Neighborhoods, poverty, and children's well-
being: A review. Santa Monica: Rand Corporation.
Pecora, P. J., Whittaker, J. K., Maluccio, A. N., Barth, R. P. with Poltnick, R. D. (2000).
The child welfare challenge: Policy, practice and research (2n Ed.). New York:
Aldine de Gruyter.
Pelton, L. (1982). Personalistic attributions and client perspectives in child welfare:
Implications for service delivery. In T. A. Wills (Ed.), Basic processes in helping.
New York: Academic Press.
Pelton, L. (1998). Four commentaries: How we can better protect children from abuse
and neglect. The Future of Children, 5(1), 126-129.
178
Peterson, R.D., Krivo, L.J. & Harris, M.A. Disadvantage and neighborhood violent
crime: Do local institutions matter? Journal of Research in Crime and
Delinquency, 37 (1), 31-63.
Pierce, G., Sarason, B., Sarason, I., Joseph, H., and Henderson, C. (1996).
Conceptualizing and assessing social support in the context of the family. In
Pierce, G., Sarason, B., and Sarason, I. (Eds.), Handbook of social support and
the family (pp. 3-24). Plenum Press, New York,
Pinderhughes, E.E., Nix, R., Foster, E.M., Jones, D. (2001, Nov.) Parenting in Context:
Impact of Neighborhood Poverty, Residential Stability, Public Services, Social
Networks, and Danger on Parental Behaviors. Journal of Marriage and the
Family, 63 (4), 941-953.
Polansky, N.A., Ammons, P.W. & Gaudin, J.M. (1985). Loneliness and isolation in child
neglect. Social Casework: The Journal of Contemporary Social Work, 6, 38^t7.
Polansky, N.A., Chalmers, M.A., Buttenwieser, E. & Williams, D.P., (1981). Damaged
parents. The University of Chicago Press, Chicago, IL.
Powell, D.R. (1994). Evaluating family support programs: Are we making progress? In
S.L. Kagan & B. Weissbound (Eds.). Putting families first (pp. 441-470). San
Francisco: Jossey-Bass Publishers.
Promising Practices Network (2002). Nurse family partnership. Retrieved November 20,
2005, from the Promising Practices Network web site:
http://www.promisingpractices.net/program.asp?programid= 16
Racino, J. A. (1998). Innovations in family support: what are we learning? Journal of
Child and Family Studies, 7(4), 433-449.
179
Rutter, M. (1995). Clinical implications of attachment concepts: Retrospect and prospect.
Journal of Child Psychology and Psychiatry, 36 (4), 549-571.
Sampson, R.J., Morenoff, J.D. & Gannon-Rowley, T. (2002). Assessing "neighborhood
effects": Social processes and new directions in research. Annual Review of
Sociology, 28, 443^78.
Schene, P. (2001). Examples of differential response in several states. National Child
Welfare Research Center for Family-Centered Practice. Retrieved online April 20,
2005 from: http://cwresource.org
Schuerman, J.R., Rzepnicki, T.L. & Littell, J.H. (1994). Putting families first: An
experiment in family preservation. New York: Aldine de Gruyer.
Shinn, M & Toohey, S.M. (2003). Community contexts of human welfare. Annual
Review of Psychology, 54, 421-459.
Simons, J.A., Irwin, D.B., & Drinnien, B.A. (1987). Psychology-the search for
understanding. New York: West Publishing Company.
Small, M.L. (2006). Neighborhood institutions as resource brokers: Childcare centers,
interorganizational ties, and resource access among the poor. Social Problems, 53
(2), 274-292.
Small, M.L. & Stark, L. (2005). Are poor neighborhoods resource deprived? A case
study of childcare centers in New York. Social Science Quarterly, 86, 1013-1036.
Socolar, R.R.S., Runyan, D.K., Amaya-Jackson, L. (1995). Methodological and ethical
issues related to studying child maltreatment. Journal of Family Issues, 16 (5),
565-586.
180
Soper, B., Milford, G.E., Rosenthal, G.T. (1995) Belief when evidence does not support
theory, Psychology and Marketing, 12 (5), 415-422.
Spruance, S.L., Reid, J.E., Grace, M., Samore, M. (August 2004). Hazard ratio in clinical
trials. Antimicrobial Agents and Chemotherapy, 48 (8), 2787-2792.
Sweet, M. A. & Appelbaum, M.I. (2004). Is home visiting an effective strategy? A meta-
analytic review of home visiting programs for families with young children. Child
Development, 75 (5), 1435-1456.
Thoits, P.A. (1986) Social support as coping assistance. Journal of Consulting and
Clinical Psychology 54, 416-423.
Thoits, P.A. (1995). Stress, coping and social support processes: Where are we? What
next? Journal of Health and Social Behavior, 35 (extra issue), 53-79.
Thomas, K. L., Berrick, J.D., & Koren, M.K. (N.D.). Voluntary placement and the Title
IV-E Waiver: Implementing reform in a changing child welfare context. Center
for Social Services Research, Unpublished document.
Tienda, M. (1991). Poor people and poor places: Deciphering neighborhood effects on
poverty outcomes. In J. Haber (Ed.), Macro-micro linkages in sociology (pp. 244-
262). Newbury Park: Sage Publications.
Tilbury, C. (2005). Counting family support. Child and Family Social Work, 10, 149—
157.
Turner, RJ . & Avison, W.R. (Nov., 1985). Assessing risk factors for problem parenting:
The significance of social support. Journal of Marriage and the Family, 47 (4)
881-892.
181
U.S. Department of Health and Human Services, Administration on Children, Youth and
Families (2007). Child Maltreatment 2005. Washington, DC: U.S. Government
Printing Office.
US Department of Health and Human Services, Administration for Children and
Families/Children's Bureau and Office of the Assistant Secretary for Planning and
Evaluation. (2005). Alternative responses to child maltreatment: Findings from
NCANDS. Washington, DC: U.S. Government Printing Office. Retrieved online
October 15, 2007 from the Office of the Assistant Secretary for Planning ad
Evaluation web site: http://aspe.hhs.gov/hsp/05/child-maltreat-resp/
van Ijzcndoorn, M.H., Juffer, F. & Duyvesteyn, M.G.C. (1995). Breaking the
intergenerational cycle of insecure attachment: A review of the effects of
attachment-based interventions on maternal sensitivity and infant security.
Journal of Child Psychology and Psychiatry 36, 225-248.
Virginia Department of Social Services. (2003). Evaluation of the Differential Response
System. Retrieved March 9, 2005 from http://www.bsc-
ca.org/Lists/Weekly%20Highlights/Attachments/62/EvalutationOutcomes_VA_D
R.pdf
Virginia Department of Social Services (2004). Evaluation of the differential response
system. Retrieved online April 20, 2005 from the Virginia Department of Social
Services web site:
http://www.dss.virginia.gov/files/division/dfs/cps/reports/eval_drs_2004.pdf
182
Wagner, M., Spiker, D., Linn, M.I., Gerlach-Downie, S., & Hernandez, F. (2003).
Dimensions of parental engagement in home visiting programs: Exploratory
study. Topics in Early Childhood Special Education, 23 (4), 171-187.
Wahba, M.A. & Bridwell, L.G. (1976). Maslow reconsidered: A review of the research
on the need hierarchy theory. Organizational Behavior and Human Performance,
15, 212-240.
Waldfogel, J. (1998a). The future of child protection: How to break the cycle of abuse
and neglect. Cambridge: Harvard University Press.
Waldfogel, J. (1998b). Rethinking the paradigm for child protection. The Future of
Children, 8 (1), 104-119.
Waldfogel, J. (2000). Reforming child protective services. Child Welfare, 79 (1), 43-57.
Wasik, B. H. (1993). Staffing issues for home visiting programs. The Future of Children,
3(3), 140-157.
Wasik, B. H. & Roberts, R.N. (1994). Home visitor characteristics, training, and
supervision: Results of a national survey. Family Relations, 43, 336-341.
Wilson, W.J. (1987). The Truly Disadvantaged: The Inner City, the Underclass, and
Public Policy. Chicago, IL: University of Chicago Press.
Wilson, W.J. (1996). When work disappears: The world of the new urban poor. New
York: Knopf.
Winefield, H.R. (1987). Psychotherapy and social support: Parallels and differences in
the helping process. Clinical Psychology Review, 7,631-64.
183
Wolock, I., Sherman, P., Feldman, L.H., & Metzger, B. (2001). Child abuse and neglect
referral patterns: A longitudinal study. Children & Youth Services Review, 23 (1),
21-47.
Yatchmenoff, D. K. (2005). Measuring client engagement from the client's perspective in
nonvoluntary child protective services. Research on Social Work Practice, 15 (2),
84-96.
Yoshikama, R. T. & Emlen, A.C. (1983). A comparison of voluntary and court-ordered
foster care: Decisions services, and parental choice. Washington, D.C.: The
Children's Bureau, Administration for Children, Youth, and Families, Office of
Human Development Services, Department of Health and Human Services.
Young, L. (1964). Wednesday's children: A study of child neglect and abuse. New York:
McGraw-Hill.
184
Figure 1: ARS Program Diagram
Allegations of Child Abuse/Neglect
CPS Hotline
Screening
- ! •
i r
DCFS Liaison
CBO Liaison
t
Screened out cases in
West Oakland, East
Oakland, and South
Havward zin codes
In-Home Assessment and Triage
Moderate to High Risk
Intensive Family Support
• Maximum 9 months
•
•
1:13 CJa
Linkag
ise Manager: family
es to other service pn
ratio
Mders
Low Risk
Community
Referrals
Short-term outcomes
Figure 2: Another Road to Safety Logic Model
Intermediate outcomes Long-term outcomes
Ensure that basic family needs are met
Promote attachment & bonding in parent-
child relationships
Provide social support
Connect families with institutional
Family isolation is decreased
Medical, dental, and nutritional needs
are met
Families understand how to navigate the
system
Improved parent-child interactions
Family stress is decreased
Parents have an improved understanding
of their rights and responsibilities
The home is a safer place
Families are integrated into their
communities
Families achieve economic self-
sufficiency and elimination of
the need for outside intervention
Improved child social,
developmental, and emotional
well-being
Improved school readiness
Families do not come into
further contact with the child
welfare system
Figure 3: Model of stress, coping, and outcomes
Stress:
Illegal status
Poverty/difficulties meeting basic
needs
Depression
Domestic violence
Loss of job
Family resources
"Individual and collective
strengths at the time the
stressor event occurs.
Examples are economic
security, health,
intelligence, job
skills..social supports"
(Boss, 1987, p. 702)
Outcomes:
Family adaptation
or crisis
Moderators Social support
Emotional,
instrumental, and
informational social
support
Figure 4: Sampling, data collection & analysis for qualitative study
Sample
Data collection
Qualitative study
X Non-probability accidental sample: All
ARS clients receiving ARS services
agree to participate (n=34)
Closed-ended telephone interviews
guided by a script
Analysis
1 ARS staff: Administrators from
community-based organizations, ECC,
SSA (n=12) and home visitors (n=10)
X Open-ended in-person interviews
guided by a script
Focus groups with administrators and
home visitors, guided by a script.
Coding for emergent themes, based on
grounded theory
Coding for emergent themes, based on
grounded theory
Figure 5: Sampling, data collection & analysis for outcomes study
Outcomes study
J :
Sample
Treatment group:
Clients referred to ARS-South Hayward who were retained for
services, siblings removed (n=161)
Comparison group:
Clients eligible for ARS services in South Hayward, but not
referred, siblings removed (n=477)
Administrative data on child welfare involvement subsequent to
ARS referral
Administrative data on child welfare involvement subsequent to
index report
Data collection
Analysis
Comparison of outcomes with comparison groups, using survival
analysis
189
Figure 6: Map of Alameda County, with ARS zip codes marked
Alameda County
j West Oakland (Prescott Joseph) |
190
Figure 7: 1 mile buffers, census tracts and zip codes
Buffers, census tracts Buffers, zip codes
191
Figure 8: Zip codes—Need for & availability of services
* j
«£'£ 4,.--
Ifei".
fc
Legend
| Low ifici'dencE
rfecijym incicterif.
-Ii.jn in-irlonrr
I .OJiLMUl
Low Availability
Medium Availability
High Availability
192
Figure 9: Census tracts—Need for & availability of services
Legend
; Low itcidBnct
-̂••••'Tjj Medium incidenci
MM High incidence
1 .cue i id
Low Availability
Medium Availability
High Availability
Figure 10: Re-report as failure
Log-rank test for equality of survivor functions
Treatment 0 1 TOTAL
Events observed 163 48 211
Events expected 164.3 46.7 211
P-value=0.83
Kaplan-Meier Survival Estimates by treatment Re-report as failure
500 1000 Days
1500
95% CI
t x = 0
95% CI tx= 1
2000
Estimated hazard functions by treatment Re-report as failure
1000 Days
1500
95% CI 95% CI
tx = 0 tx = 1
Figure 11: Investigation as failure
Log-rank test for equality of survivor functions
Treatment 0 1 TOTAL
Events observed 100 38 138
Events expected 107.6 30.4 138
P-value=0.12
Kaplan-Meier Survival Estimates by treatment Investigation as failure
500 1000 Days
1500
95% CI
tx = 0
95% CI
tx = 1
2000
Estimated hazard functions by treatment Investigation as failure
500 1000 Days
95% CI 95% CI
tx = 0 tx= 1
Outcome as failure: families with a re-report only
Log-rank test for equality of survivor functions
P-value=0.002
195
Treatment 0 1 TOTAL
Events observed 100 38 138
Events expected 113.8 24.2 138
Kaplan-Meier Survival Estimates by tx, families with re-report Investigation as failure
500 1000 Days
1500
95% CI 95% CI tx = 0 tx= 1
2000
Estimated hazard functions by tx, families with re-report Investigation as failure
1000 Days
1500 2000
95% CI
tx = 0
95% CI
tx= 1
Figure 12: Substantiation as failure
Log-rank test for equality of survivor functions
Treatment 0 1 TOTAL
Events observed 18 9 27
Events expected 21 6 27
P-value= 0.1657
1000 Days
95% CI
tx = 0
95% CI
tx = 1
1 ^ •
in .
m <\i -
o -
Kaplan-Meier Survival Estimates by treatment Substantiation as failure
Estimated hazard functions by treatment Substantiation as failure
1000 Days
95% CI
tx = 0
1500
95% CI
tx= 1
2000
Substantiation as failure: families with an investigation only
Treatment 0 1 TOTAL
Events observed 18 9 27
Events expected 20.6 6.4 27
P-value=0.23
Kaplan-Meier Survival Estimates by tx, families with investigation Substantiation as failure
1000 Days
95% CI
tx = 0
1500 2000
95% CI tx= 1
Estimated hazard functions by tx, families with investigation Substantiation as failure
500 1000 Days
1500
95% CI
tx = 0
95% CI
tx= 1
2000
Table 1: State implementation of Differential Response, structural elements
State
North Carolina
Missouri
Minnesota
Virginia
Washington
Paraprofessionals
No, cases are managed by
CPS social workers
No, cases are managed by
CPS social workers
In most cases, no. In the
majority of counties, CPS
social worker acts as case
manager.
No, cases are managed by
CPS social workers
No, community-based
organization staffed by
Masters level social work
professionals and public
health nurses.
Duration of
services
Not specified
Not specified
In most
counties, up to
90 days from
the conclusion
of the
assessment
45-60 days
Not specified
Use of home
visitation
Yes, case
management with
weekly home visits
No, case
management
specified but not
home visitation
No, case
management
specified but not
home visitation.
No, case
management
specified but not
home visitation.
Yes (frequency not
given)
Table 2: Institutional resources dimensions examined, by study
Study Peterson, Krivo & Harris, 2000 Katz, Kling, Liebman, 2001
Klebanov, etal., 1998
Brooks-Gunn, et al„ 1998 Ludwig & Ladd, 1997 Chase-Lansdale et al., 1997 Ensminger, Lamkin & Jacobson, 1996 Fuller et al, 1997 Small & Stark, 2005 Fuller et al., 2004 Levanthal & Brooks-Gunn, 2000 (Gautraux study)
Resource type Recreation centers
Playgrounds
Learning, social, and recreational opportunities Medical facilities
Educational facilities Learning activities
Educational facilities
Child care
Child care
Child care
Employment opportunities for youth
Availability • /
X
/
X
X
r
s*
</
s X
•/
Accessibility X
X
X
X
X
X
X
X
X
X
X
Affordability X
X
X
X
X
X
X
X
X
X
X
Quality X
•*
X
/ *
V
X
X
X
X
V
s
Study addressed this dimension
Study did not address this dimension
Authors speculated on, but did not assess, this dimension
Table 3: Data on service types (total services=7,952)
Alcohol and drug treatment
Alcoholics Anonymous & Narcotics Anonymous meeting sites
Basic needs social services
Child care
Churches
Dental programs (publicly funded)
Employment social services
Health (agencies and school-based programs)
Hospitals
Immigrant social services
Legal social services
Libraries
Medical facilities
Mental health (agencies and school-based programs)
Youth development social services
Table 4: Availability of services in Alameda County, by zip code
Geographic area
Within zip codes Within zip codes and 1 mile buffers Within ARS zips Within ARS zip codes and 1 mile buffers
Minimum
46 55
841.00 1017
Maximum
3305 3445
2374.00 2756
Mean
1027 1172
1415 1681
Standard Deviation 622 681
492 551
Table 5: Comparison of demographic data, treatment and comparison groups
GENDER
Tx
Notx
Male 50% N=80 55% N=282
Female 50% N=80 45% N=229
PRIMARY ETHNICITY
Tx
Notx
Black
11% N=19 21% N=109
White
18% N=29 24% N=121
Hispanic
28% N=44 39% N=199
Ethnicity-Other 16% N=25 10% N=49
Ethnicity-Unknown 27% N=43 6% N=33
AGE AT TIME OF ARS REFERRAL OR INDEX REPORT
Tx
No tx
During pregnancy or at birth 1% N=2 0% N=l
Infancy (birth to age 2) 38% N=61 28% N=142
Toddler (2-3)
15% N=24 16% N=84
Preschool (3-5)
45% N=72 55% N=281
Elementary age (6 to 12)** 1% N=l 1% N=3
NUMBER OF CHILD MALTREATMENT REPORTS PRIOR TO ARS REFERRAL OR INDEX REPORT
Tx
No tx
Prior reports (none) 10% N=16 34% N=175
Prior reports (1 or more) 90% N=144 66% N=336
RISK LEVEL FOR TREATMENT GROUP
Risk level Missing
Low
Moderate
High
Very High
Treatment group 18% N=29 2% N=3 40% N=64 33% N=52 8% N=12
Note: Percentages are rounded and may not sum to 100
**Of the children in the elementary age category, all three in the non-treatment group were six years old and the one in the treatment group was 12.
Table 6: Comparison of clients with re-report, investigated re-report, and substantiated re-report, treatment and comparison groups
RE-REPORTS
Tx
No tx
Total number of re-reports
30% N=48 32% N=163
Maximum time to re-report 1765 days
1946 days
Minimum time to re-report 295 days
274 days
Mean time to re-report
641.8 days
755.1
Standard deviation
334.5
395.8
RE-REPORTS: ALLEGATION TYPES
Emotional abuse
Neglect/ caretaker absence or incapacity*
Physical abuse
Sexual abuse
Substantial risk
Not determined
Tx 6% N=3
37% N=18
31% N=15
21% N=10
4% N=2
0% N=0
No tx
11% N=18
43% N=70
31% N=50
15% N=24
0% N=0
1% N=l
*T Includes severe neglect
INVESTIGATION: RESPONSE TYPES
Tx
Notx
Immediate 58% N=22 66% N=66
Ten day 42% N=16 44% N=44
4VESTIG
Tx
Notx
ATION: CONC1 Inconclusive 21% N=8 19% N=19
LUSION TYPES Substantiated 24% N=9 18% N=18
Unfounded 50% N=19 58% N=58
Not listed 5% N=2 5% N=5
SUBS' ANTIATION: ALLEGAT ON TYPES Emotional abuse
Neglect/ caretaker absence or incapacity*
Physical abuse
Sexual abuse
Substantial risk
Not determined
Tx n/a 33% N=3
11% N=l
33% N=3
22% N=2
n/a
No tx n/a
61% N=3
22% N=4
17% N=3
n/a n/a
*Includes severe neglect Note: Percentages are rounded and may not sum to 100
Top Related